Uploaded byrevathi148366
PPTX, PDF178 views

artificialintelligencedata driven analytics23.pptx

data

Embed presentation

Download to read offline
Artificial Intelligence
Introduction 2  What is AI?  Introduction to AI Levels?  Types of Artificial Intelligence  AI VS machine learning vs deep learning  Where is AI used?  AI use cases  Why is AI booming now?  AI trend in 2020 01
Artificial Intelligence Transforming the Nature of Work, Learning, and Learning to Work Artificial intelligence (AI) is a popular branch of computer science that concerns with building “intelligent” smart machines capable of performing intelligent tasks. With rapid advancements in deep learning and machine learning, the tech industry is transforming radically. Artificial Intelligence Machine Learning . Deep Learning 3
Introduction to AI Levels? Types of Artificial Intelligence Artificial Narrow Intelligence 4 Artificial Super Intelligence Artificial General Intelligence
Artificial Intelligence 2020 2019 2018 65% . 35% . 2020 2019 2018 AI AI 2016 2017 2018 2019 2020 Artificial intelligence (AI) is a popular branch of computer science that concerns with building “intelligent” smart machines capable of performing intelligent tasks. 6 With rapid advancements in deep learning and machine learning, tech industry is transforming radically.
Machine Learning Machine learning is a type of AI that enables machines to learn from data and deliver predictive models. The machine learning is not dependent on any explicit programming but the data fed into it. It is a complicated process. Based on the data you feed into machine learning algorithm and the training given to it, an output is delivered. A predictive algorithm will create a predictive model. This slide is 100% editable. Adapt it to your needs and capture your audience's attention. 7 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
Deep Learning Deep Learning is a subfield of machine learning that is concerned with algorithms inspired by the brain's structure & functions known as artificial neural networks A computer model can be taught using Deep Learning to run classification actions using pictures, texts or sounds as input 8
Where is AI used? Real-time Operations Management Customer Services Risk Management & Analytics Customer Insight Pricing & Promotion Predictive Analytics Customer Experience Supply Chain Knowledge Creation Research & Development Fraud Detection Human Resources 9 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
AI Usecase in HealthCare AI and Robotics Training Research 10 This slide is 100% editable. Adapt it to your needs and capture your audience's attention. Early Detection Keeping Well End of Life Care Treatment Decision Making Diagnosis
AI Use Cases in Human Resource Recruiting Dynamic Career Sites Smart Sourcing Onboarding Automated Messages Curated Videos Learning Curated Training Skill Development Engagement HR Chatbot Engagement Surveys 11 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
AI in Banking for Fraud Detection Cardholder Profiles Postings Payment System Nonmonetary System Analyst Workstation Payment and Non-Monetary Transactions 8 Authorization System Rules Definition Configuration Workstation Case Management Database Case Creation Module Neural Network Engine Scoring Engine Expert Authorization Response Module Expert Rules Base 6 Case Creation Rules Execute 1 Auth Request Expert Rules Execute 3 4 Auth Recommendation 22 Auth Request & Score 5 Transaction & Score 7 Case Information 12 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
AI in Supply Chain 16 Structured & Unstructured Information Regulator y Data B2B Transactio n Data Inventor y Data Multimedi a Data Sensor Data Logistics Data Trading Partner Data Social Medi a Data Digital Ecosystem Data Lake Pervasive Visibility Proactive Replenishment Predictive Maintenance Secure Device Maintenance Ecosystem Integration Unified Messaging Actionable Insights IIOT – Securely Provisionally People, System and Things Secure Access Via Identity Management for Transient Users Procurement Manufacturing Customers Service Logistics
Ai Chatbots in Healthcare Search Engine Users learn to search for information Social Platforms Like Facebook connect users online 14 Smartphones Bring the internet online Healthbots Bring all of the above together for healthcare use cases Artificial Intelligence Self Learning machines becomes smarter the more they are used Messenger Apps Lets users chat anywhere, anytime App Eco-system Lets users download and use apps easily
Why is AI booming now? 8,097 7,540 7,336 4,680 4,201 3,714 3,655 3,564 3,169 2,000 0 4,000 6,000 8,000 Static Image Recognition, Classification and Tagging Algorithm Trading Strategy Performance Management Efficient, Scalable Processing of Patient Data Predictive Maintenance Object Identification, Text Query of Images detection, classification, tracking Automated Geophysical Content Distribution on Feature Detection Social Media Object detection & classification, avoidance, navigation 24% 22% 27% 37% 36% 51% 60% 60% 66% 71% 0 20 40 60 80 100 HR/Workforce Management Operational Environment Monitoring Through External Devices/Systems Logistics & Supply Chain Production Floor Systems Expediting Transactions Fleet Mobile External Communication Customer Relation/Interaction (i.e., chatbots) Security/Fraud Global AI Revenue Forecast by 2025, Ranked by Use Case in millions US Dollar Penetration of Artificial Intelligence Skills, by Country Organizations deploying AI, by Functional Areas Unite d States 100% 92% China 84% Indi a 54% Israe l 45% German y 15
10 AI Trend in 2020 Robotic Process Automation AI will make Healthcare more Accurate Data Modeling will move to the Edge AI will Come for B2B Ai-powered Chatbots AI In Retail Aerospace and Flight Operations Controlled by AI AI Mediated Media and Entertainment Advanced Cybersecurity Automated Business Process “2020 AI Trends” 16
Machine Learning 02  What is Machine Learning?  7 Steps of Machine learning  Machine Learning vs. Traditional Programming  How does machine learning work?  Machine learning Algorithms  Machine learning use cases  How to choose Machine Learning Algorithm  Why to use decision tree algorithm learning  Challenges and Limitations of Machine learning  Application of Machine learning  Why is machine learning important? 20
Machine Learning Traditional Programming Data (Input ) Program Output Data (Input ) Output Machine Learning Program Machine Learning is the result of General AI that involves developing machines that can deliver results better than humans Input Data: Feed Learner Various Data Output Data: Present Rules “Learning” Machine Learning System 18
7 Steps of Machine Learning 01 02 03 04 05 06 07 Choosing a Model 19 Training Preparing that Data Evaluation Prediction Gathering Data Hyperparameter Tuning
Machine Learning vs. Traditional Programming Prediction Result Compute r Dat a Handcrafte d Model Learning Mode l Prediction Result New Data Mode l Sample Data Expected Result Compute r Compute r Traditional Modelling Machine Learning 20
How does Machine Learning Work? Collect Data Collect data from hospitals, health insurance companies, social service agencies, police and fire dept. Define Objectives Identify, the problem to be solved and create a clear objective. 21 Prepare Data Preparing data is a crucial step and involves building workflows to clean, match and blend the data. Select Algorithm Depend on the problem to be solved and the type of data an appropriate algorithm will be chosen. Train Model Data is fed as input and the algorithm configured with the required parameters. A percent of the data can be utilized to train the model. Integrate Model Publish the prepared experiment as a web service, so applications can use the model. Test Model The remaining data is utilized to test the model, for accuracy. Depending on the results, improvements, can be performed in the “Train model’ and/or “Select Algorithm” phases, iteratively.
Machine Learning Algorithms  KNN  Trees  Logistic Regression  Naïve-Bayes  SVM Reinforcement Categorical Machine Learning  Apriori  FP-Growth Associatio n Anlaysis Hidden Markov Model  SVD  PCA  K-means Clusterin g Unsupervised Regression  Linear  Polynomial Decision Tree Classificatio n Supervised Random Forest Continuous 22
Machine Learning Use Cases Energy Feedstock & Utilities  Power Usage Analytics  Seismic Data Processing  Your Text Here  Smart Grid Management  Energy Demand & Supply Optimization Financial Services  Risk Analytics & Regulation  Customer Segmentation  Your Text Here  Credit Worthiness Evaluation Travel & Hospitality  Aircraft Scheduling  Dynamic Pricing  Your Text Here  Traffic Patterns & Congestion Management Manufacturing  Predictive Maintenance or Condition Monitoring  Your Text Here  Demand Forecasting  Process Optimization  Telematics Retail  Predictive Inventory Planning  Recommendatio n Engines  Your Text Here  Customer ROI & Lifetime Value Healthcare & Life Sciences 23  Alerts & Diagnostics from Real-time Patient Data  Your Text Here  Predictive Health Management  Healthcare Provider Sentiment Analysis
How to Choose Machine Learning Algorithm What do you want to do with your Data? Algorithm Cheat Sheet Additional Requirements Accuracy Linearity Number of Parameters Training Time Number of Features How to Select Machine Learning Algorithms 24
Why use Decision Tree Machine Learning Algorithm? To Classify Non-linear Relationship between Predictors & Response Linear Relationship between Predictors & Response Use c4.5 Implementatio n Use Standard Regression Tree Responsible Variable has only 2 Categories Use Standard Classification here Use c4.5 Implementatio n To Predict Responsible variable is Continuous Decision Trees 25 Response Variable has Multiple Categories
Challenges and Limitations of Machine learning Advantages Easily Identifies Trends and Patterns No Human Intervention needed Handling multi- dimensional & multi- variety Data Continuous Improvement Wide Applications Data Acquisitio n Time and Resource s High error- Susceptibilit y Interpretatio n Results Disadvantages 26
Application of Machine Learning Automatic Language Translation Medical Diagnosis Stock Market Trading Online Fraud Detection Virtual Personal Assistant Email Spam and Malware Filtering Self Driving Cars Product Recommendations Traffic Prediction Speech Recognition Image Recognition 27
Why is Machine Learning Important? Phase 1 : Learning Training Data Phase 2: Prediction  Precision/ recall  Over fitting  Test/cross Validation data, etc. Error Analysis  Normalization  Dimension Reduction  Image Processing, etc. Pre-Processing  Supervised  Unsupervised  Minimization, etc. Learning Predicted Data 28 Prediction New Data Model
Deep Learning 03  What is Deep Learning?  Deep learning Process  Classification of Neural Networks  Types of Deep Learning Networks  Feed-forward neural networks  Recurrent neural networks (RNNs)  Convolutional neural networks (CNN)  Reinforcement Learning  Examples of deep learning applications  Why is Deep Learning Important?  Limitations of deep learning 32
What is Deep Learning? Deep Learning is a subfield of machine learning that is concerned with algorithms inspired by the brain's structure and functions known as artificial neural networks. A computer model can be taught using Deep Learning to run classification actions using pictures, texts or sounds as input. Car Not Car Output Input Feature Extraction + Classification What is Deep Learning? 31
Deep Learning Process Understand the Problem This slide is 100% editable. Adapt it to your needs and capture your audience's attention. 32 Identify Data This slide is 100% editable. Adapt it to your needs and capture your audience's attention. Select Deep Learning Algorithms This slide is 100% editable. Adapt it to your needs and capture your audience's attention. Training the Model This slide is 100% editable. Adapt it to your needs and capture your audience's attention. Test the Model This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
Classification of Neural Networks Input Layer Hidden Layer Output Layer x1 x2 xn v11 v12 vpn w11 w22 wmp ym y2 y1 V1n 1 2 p w1p 1 33 2 m
Types of Deep Learning Networks Artificial Neural Networks Used for Regression & Classification Convolutional Neural Networks Used for Computer Vision Supervised Recurrent Neutral Networks Used for Time Series Analysis Self-Organizing Maps Used for Feature Detection Deep Boltzmann Machines Used for Recommendation Systems Unsupervised AutoEncoders Used for Recommendation Systems Deep Learning Models Supervised  Artificial Neural Networks (ANN)  Convolutional Neural Networks (CNN)  Recurrent Neural Networks (RNN) Unsupervised  Self Organizing Maps (SOM)  Boltzmann Machines (BM)  AutoEncoders (AE) 34
Feed-forward Neural Networks Input Layer Hidden Layer Output Layer Variable- #1 35 Variable- #2 Variable- #3 Variable- # 4 Output
Recurrent Neural Networks (RNNs) x1 x2 Input Layer 36 Recurrent Network Output Layer Hidden Layers
Convolutional Neural Networks (CNN) Take Car = A1 37 Truck = B1 VAN = C1 Bicycle = D1 Rest all be Same
Reinforcement Learning State, Reward Action Agent Environment Exploration Policy Neural Networks Filters Memor y Algorith m Reinforcement Learning uses rewards and punishment to train computing models to perform a sequence of selections. Here computing faces a game-like scenario where it employs trial and error to answer. Based on the action it performs, computing gets either rewards or penalties. Its goal is to maximize the rewards. 38
Examples of Deep Learning Applications Image Recognitio n Natural Language Processing Portfolio Management & Prediction of Stock Price Movements Drug Discovery & Better Diagnostics of Diseases in Healthcare Speech Recognitio n Robots and Self - Driving Cars 39
Why is Deep Learning Important? Performance Data Deep Learning Other Learning Algorithms 40
Limitations of Deep Learning Limitations of Deep Learning Amount of Data Statistical Reasoning 41 Interpretability
Difference between AI vs ML vs DL 04  What is AI?  What is ML?  What is Deep Learning?  Machine Learning Process  Deep Learning Process  Difference between Machine Learning and Deep Learning  Which is better to start AI,ML or Deep learning 44
Difference between AI vs ML vs DL Machine Learning Ability to learn without being explicitly programmed Deep Learnin g Learning based on deep neural network Artificial Intelligence Engineering of making intelligent machines and programs 43 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
What is AI? Artificial Intelligence Artificial Intelligence With rapid advancements in deep learning and machine learning, the tech industry is transforming radically. (AI) is a popular branch of computer science that concerns with building “intelligent” smart machines capable of performing intelligent tasks. 44 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
What is ML? Learns Predicts Improves Machine Learning Ordinary System With AI Machine Learning is a type of AI that enables machines to learn from data and deliver predictive models. The machine learning is not dependent on any explicit programming but the data fed into it. It is a complicated process. Based on the data you feed into machine learning algorithm and the training given to it, an output is delivered. A predictive algorithm will create a predictive model. Introduction to Machine learning 45
What is Deep Learning? Artificial intelligence (AI) is a popular branch of computer science that concerns with building “intelligent” smart machines capable of performing intelligent tasks. With rapid advancements in deep learning and machine learning, the tech industry is transforming radically. 46 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
Machine Learning Process Data Raw & Training Data Modelling Candidate & Final Visualisation Predictions & Strategy Data Gathering Data Cleaning Selecting Right Algorithms Building Model & Finalising Data Transformation into Predictions 47 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
Deep Learning Process Test the Model 48 Understand the Problem Identify Data Select Deep Learning Algorithm Training the Model
Difference between Machine Learning and Deep Learning Inpu t Feature Extractio n Classifica -tion Outpu t Car Not Car Inpu t Feature Extraction + Classification Outpu t Car Not Car 49 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
Which is better to start AI,ML or DL? Artificial Intelligence Any Technique which enables computers to mimic human behavior. 50 This slide is 100% editable. Adapt it to your needs and capture your audience's attention. Machine Learning Subset of AI Techniques which use Statistical Methods to Enable Machines to Improve with Experiences. Deep Learning Subset of ML which make the Computation of Multi- layer Neural Networks Feasible. 01 02 03
Supervised Machine Learning 05  Types of Machine Learning  What is Supervised Machine Learning?  How Supervised Learning Works  Types of Supervised Machine Learning Algorithms  Supervised vs. Unsupervised Machine learning techniques  Advantages of Supervised Learning:  Disadvantages of Supervised Learning 53
Types of Machine Learning Inputs Training Inputs Makes Machine Learn Explicitly Data with Clearly defined Output is given Direct feedback is given Predicts outcome/future Resolves Classification and Regression Problems Supervised Learning Unsupervised Learning Machine Understands the data (Identifies Patterns/ Structures) Evaluation is Qualitative or Indirect Does not Predict/Find anything Specific Reinforcement Learning An approach to AI Reward Based Learning Learning form +ve & +ve Reinforcement Machine Learns how to act in a Certain Environment To Maximize Rewards Inputs Rewards Outputs 52 This slide is 100% editable. Adapt it to your needs and capture your audience's attention. Outputs Outputs
What is Supervised Machine Learning? Input Raw Data Output Processing Algorithm Training Data set Desired Output Supervisor Supervised Learning 53 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
How Supervised Machine Learning works Classification Sorting Items into Categories Regression Identifying Real Values (Dollars, Weight, etc.) Label “Group 1” Machine Step1 Provide the Machine Learning Algorithm Categorized or “labeled” Input and Output Data from to Learn Group 1 Group 2 Mach ine Step2 Feed the Machine New, Unlabeled Information to See if it Tags New Data Appropriately. If not, Continue Refining the Algorithm Types of Problems to which it’s Suited 54 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
Types of Supervised Machine Learning Algorithms Classification  Fraud Detection  Email Spam Detection  Diagnostics  Image Classification. Regression  Risk Assessment  Score Prediction Supervised Learning 55 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
Supervised vs. Unsupervised Machine Learning Techniques V S Supervised Learning Input & Output Data  Classification  Regression Predictions & Predictive Models Unsupervised Learning Input Data  Clustering  Association Patterns / Structure Discovery 56 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
It allows you to be very specific about the definition of the labels. In other words, you'll train the algorithm to differentiate different classes where you'll set a perfect decision boundary. You are ready to determine the amount of classes you would like to possess. 57 This slide is 100% editable. Adapt it to your needs and capture your audience's attention. The input file is extremely documented and is labeled. The results produced by the supervised method are more accurate and reliable as compared to the results produced by the unsupervised techniques of machine learning. this is often mainly because the input file within the supervised algorithm is documented and labeled. this is often a key difference between supervised and unsupervised learning. The answers within the analysis and therefore the output of your algorithm are likely to be known thanks to that each one the classes used are known. Advantages of Supervised Learning Advantages
Disadvantages of Supervised Learning Supervised learning are often a posh method as compared with the unsupervised method. The key reason is that you simply need to understand alright and label the inputs in supervised learning. 58 This slide is 100% editable. Adapt it to your needs and capture your audience's attention. It doesn’t happen in real time while the unsupervised learning is about the important time. this is often also a serious difference between supervised and unsupervised learning. Supervised machine learning uses of-line analysis. It is needed tons of computation time for training. If you've got a dynamic big and growing data, you're unsure of the labels to predefine the principles. this will be a true challenge.
Unsupervised Machine Learning 06  What is Unsupervised Learning?  How Unsupervised Machine Learning works  Types of Unsupervised Learning  Disadvantages of Unsupervised Learning 61
What is Unsupervised Learning? Input Raw Data Output Algorithm Interpretation Processing  Unknown output  No Training Data Set Unsupervised Learning 60 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
How Unsupervised Machine Learning works Machine Step 1 Provide the machine learning algorithm uncategorized, unlabeled input data to see what patterns it finds Similar Group 1 Similar Group 2 Machine Step 2 Observe and learn from the patterns the machine identifies Types of Problems to Which it’s Suited Clustering Identifying similarities in groups For Example: Are there patterns in the data to indicate certain patients will respond better to this treatment than others? Anomaly Detection Identifying abnormalities in data For Example: Is a hacker intruding in our network? 61 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
Types of Unsupervised Learning Dimensionality Reduction  Text Mining  Face Recognition  Big Data Visualization  Image Recognition Clustering  Biology  City Planning  Targeted Marketing Unsupervised Learning 62 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
Disadvantages of Unsupervised Learning You cannot get very specific about the definition of the info sorting and therefore the output. This is often because the info utilized in unsupervised learning is labeled and not known. It's employment of the machine to label and group the data before determining the hidden patterns. 63 This slide is 100% editable. Adapt it to your needs and capture your audience's attention. Less accuracy of the results. This is often also because the input file isn't known and not labeled by people beforehand , which suggests that the machine will got to do that alone. The results of the analysis can't be ascertained. there's no prior knowledge within the unsupervised method of machine learning. Additionally, the numbers of classes also are not known. It results in the lack to determine the results generated by the analysis.
Reinforcement learning 07  What is Unsupervised Learning?  How Unsupervised Machine Learning works  Types of Unsupervised Learning  Disadvantages of Unsupervised Learning 66
What is Reinforcement Learning? Reinforced Response Input Input Response It’s a mang o Feedback Wrong! It’s an apple Learns Note d It’s an Apple 65 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
How Reinforcement Learning Works? Input Raw Data Rewar d State Selection of Algorith m Best Action Environment Agent Output Reinforcement Learning 66 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
Types of Reinforcement Learning Gaming Finance Sector 67 This slide is 100% editable. Adapt it to your needs and capture your audience's attention. Inventory Management Robot Navigation Manufacturing
Disadvantage of Reinforcement Learning You cannot get very specific about the definition of the info sorting and therefore the output. this is often because the info utilized in unsupervised learning is labeled and not known. it's employment of the machine to label and group the data before determining the hidden patterns. Less accuracy of the results. this is often also because the input file isn't known and not labeled by people beforehand , which suggests that the machine will got to do that alone. The results of the analysis can't be ascertained. There's no prior knowledge within the unsupervised method of machine learning. Additionally, the numbers of classes also are not known. It results in the lack to determine the results generated by the analysis Reinforcement learning as a framework is wrong in many various ways, but it's precisely this quality that creates it useful. Too much reinforcement learning can cause an overload of states which may diminish the results. Reinforcement learning isn't preferable to use for solving simple problems Reinforcement learning needs tons of knowledge and tons of computation. it's data-hungry. that's why it works rather well in video games because one can play the sport again and again and again, so getting many data seems feasible. 68 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
Back Propagation Neural Network in AI 08  Back Propagation Neural Network in AI  What is Artificial Neural Networks?  What is Backpropagation?  Why We Need Backpropagation?  What is a Feed Forward Network?  Types of Backpropagation Networks  Best practice Backpropagation 71
Back Propagation Neural Network in AI 1 2 i1 i2 h 2 w1 b1 b2 net out 70 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
What is Artificial Neural Networks? Feed-Forward Network Output Input Layer Network Inputs Hidden Layer Back Propagation Output Layer 71 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
What is Backpropagation Neural Networking? x x x w w w w Difference in Desired Values 1 Input Layer Hidden Layer(s) 1 3 Output Layer 5 Backprop Output Layer 72 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
Why We Need Backpropagation? 75 Most prominent advantages of Backpropagation Are: Backpropagation is fast, simple and straightforward to program. It has no parameters to tune aside from the numbers of input. It is a typical method that generally works well. It doesn't need any special mention of the features of the function to be learned. It is a versatile method because it doesn't require prior knowledge about the network.
What is a Feed Forward Network? Input Layer Hidden Layer Output Layer 74 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
Types of Backpropagation Networks Static Back-propagation Recurrent Backpropagation 75 This slide is 100% editable. Adapt it to your needs and capture your audience's attention. Static Back-propagation It is one quite backpropagation network which produces a mapping of a static input for static output. it's useful to unravel static classification issues like optical character recognition. Recurrent Backpropagation Recurrent backpropagation is fed forward until a hard and fast value is achieved. Then, the error is computed and propagated backward. The main difference between both of those methods is: that the mapping is rapid in static back-propagation while it's nonstatic in recurrent backpropagation
Best Practice Backpropagation A neural network is a group of connected it I/O units where each connection features a weight related to its computer programs. 76 Backpropagation is fast, simple and straightforward to program. A feedforward neural network is a man- made neural network. Backpropagation may be a short form for "backward propagation of errors." it's a typical method of coaching artificial neural networks.
Expert System in Artificial Intelligence 09  What is an Expert System?  Examples of Expert Systems  Characteristic of Expert System  Components of the expert system  Conventional System vs. Expert system  Human expert vs. expert system  Benefits of expert systems  Limitations of the expert system  Applications of expert systems 79
Types of Deep Learning Networks The Expert System in AI are computer applications. Also, with the assistance of this development, we will solve complex problems. it's level of human intelligence and expertise Knowledge Base Inference Engine User Interface User (May not be an expert) Human Expert Knowledge Engineer 78
Examples of Expert Systems The Highest Level of Expertise The expert system offers the very best level of experience. It provides efficiency, accuracy and imaginative problem-solving. Right on Time Reaction An Expert System interacts during a very reasonable period of your time with the user. the entire time must be but the time taken by an expert to urge the foremost accurate solution for an equivalent problem Good Reliability The expert system must be reliable, and it must not make any an error. Flexible It is significant that it remains flexible because it the is possessed by an Expert system. Capable of Handling Challenging Decision & Problems An expert system is capable of handling challenging decision problems and delivering solutions. Effective Mechanism Expert System must have an efficient mechanism to administer the compilation of the prevailing knowledge in it. Expert System Non- expert User Knowledge from an expert User Interface Inference Engine Knowledge Base 79 Query Advice
Characteristic of Expert System High level Performance The system must be capable of responding at A level of competency adequate to or better than an expert system within the field. the standard of the recommendation given by the system should be during a high level integrity and that the performance ratio should be also very high Domain Specificity Expert systems are typically very domain specific. For ex., a diagnostic expert system for troubleshooting computers must actually perform all the required data manipulation as a person's expert would. The developer of such a system must limit his or her scope of the system to only what's needed to unravel the target problem. Special tools or programming languages are often needed to accomplish the precise objectives of the system Good Reliability The expert system must be as reliable as a person's expert Adequate Response Time The system should be designed in such how that it's ready to perform within alittle amount of your time , like or better than the time taken by a person's expert to succeed in at a choice point. An expert system that takes a year to succeed in a choice compared to a person's expert’s time of 1 hour wouldn't be useful Understandable The system should be understandable i.e. be ready to explain the steps of reasoning while executing. The expert system should have an evidence capability almost like the reasoning ability of human experts Use Symbolic Representations Expert systems use symbolic representations for knowledge (rules, networks or frames) and perform their inference through symbolic computations that closely resemble manipulations of tongue 80
Components of the Expert System Explanation Inference Engine Knowledge Base Acquisition Facility User Interface Experts and Knowledge Engineers Users 81
Conventional System vs. Expert System vs Knowledge domain break away the mechanism processing 01 The program could have made an error 02 Not necessarily need all the input/data 03 Changes within the rule are often made with ease 04 The system can work only with the rule as a tittle 05 Information and processing combined during a sequential file 01 The program isn't wrong 02 Need all the input file 03 Changes to the program are inconvenient 04 The system works if it's complete 05 82
Human Expert vs. Expert System 85 Human Experts (Artificial ) Expert Systems Permanen t Perishabl e Easy to Transfer Difficult to Transfer Easy to Document Difficult to Document Affordable, costly to develop, but cheap to operate Expensive, especially top notch Add Your Text Here Add Your Text Here
Benefits of Expert Systems 01 Easy to Develop and Modify the System 02 Fast Response 03 Low Accessibility Cost 04 Error Rate are Very Low 05 Humans Emotions are not Affected 84
Limitations of the Expert System Don’t Have Decision Making Power Like Humans Expert System is not Widely used or Tested It cant Deal with the Mixed Knowledge There are Chances of Errors Its Difficult to Maintain Development Cost is High Not Able to Explain the Logic Behind the 85 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
Applications of Expert Systems Medical domain (Diagnosis system, medical operations Process Control System Finance/Commerce (Stock market trading, airline scheduling cargo scheduling) Warehousin g Optimizatio n Knowledge domain (Finding out the faults in vehicles, computer) Repairin g Monitorin g system Design domain (Camera lens design,automobile design) Shipping 86
Good ideas start with brainstorming Great ideas start with coffee. Coffee Break 87
Bar Chart 02 Product This graph/chart is linked to excel, and changes automatically based on data. Just left click on it and select “Edit Data”. 01 Product This graph/chart is linked to excel, and changes automatically based on data. Just left click on it and select “Edit Data”. 100 90 80 70 60 50 40 30 20 10 0 88 Jan Feb Mar Apr May Jun Sales ( in USD millions) Year 2020 100% Product 01 Product 02
Stacked Line With Markers 02 Product This graph/chart is linked to excel, and changes automatically based on data. Just left click on it and select “Edit Data”. 01 Product This graph/chart is linked to excel, and changes automatically based on data. Just left click on it and select “Edit Data”. 3628.4 89 3573.9 3484.0 3532.1 3740.3 3881.7 3528.4 3873.9 3584.0 3732.1 3640.3 3981.7 3400 3500 3600 3700 3800 3900 4000 4100 2015 2016 2017 2018 2019 2020 In Millions YEARS

More Related Content

PDF
Artificial Intelligence Machine Learning Deep Learning Ppt Powerpoint Present...
bySlideTeam
 
PDF
Artificial Intelligence PowerPoint Presentation Slide Template Complete Deck
bySlideTeam
 
PPTX
artificialintelligenceandmachinelearningpowerpointpresentationslidescompleted...
byrevathi148366
 
PPTX
demo AI ML.pptx
byPriyadharshiniG41
 
PPTX
artificialintelligencemachinelearningdeeplearningpptpowerpointpresentationsli...
bySuthahar
 
PDF
Reinforcement Learning In AI Powerpoint Presentation Slide Templates Complete...
bySlideTeam
 
PDF
Artificial Intelligence Machine Learning Deep Learning PPT PowerPoint Present...
bySlideTeam
 
PDF
Artificial Intelligence High Technology PowerPoint Presentation Slides Comple...
bySlideTeam
 
Artificial Intelligence Machine Learning Deep Learning Ppt Powerpoint Present...
bySlideTeam
 
Artificial Intelligence PowerPoint Presentation Slide Template Complete Deck
bySlideTeam
 
artificialintelligenceandmachinelearningpowerpointpresentationslidescompleted...
byrevathi148366
 
demo AI ML.pptx
byPriyadharshiniG41
 
artificialintelligencemachinelearningdeeplearningpptpowerpointpresentationsli...
bySuthahar
 
Reinforcement Learning In AI Powerpoint Presentation Slide Templates Complete...
bySlideTeam
 
Artificial Intelligence Machine Learning Deep Learning PPT PowerPoint Present...
bySlideTeam
 
Artificial Intelligence High Technology PowerPoint Presentation Slides Comple...
bySlideTeam
 

Similar to artificialintelligencedata driven analytics23.pptx

PDF
introductiontoartificialintelligencepowerpointpresentationslides-210722041732...
bySudhanshu Gupta
 
PDF
Introduction To Artificial Intelligence Powerpoint Presentation Slides
bySlideTeam
 
PDF
Artificial Intelligence And Machine Learning PowerPoint Presentation Slides C...
bySlideTeam
 
PDF
Back Propagation Neural Network In AI PowerPoint Presentation Slide Templates...
bySlideTeam
 
PDF
AI Vs ML Vs DL PowerPoint Presentation Slide Templates Complete Deck
bySlideTeam
 
PDF
What is artificial intelligence (IA) ?
byOussama Belakhdar
 
PDF
Introduction To Artificial Intelligence PowerPoint Presentation Slides
bySlideTeam
 
PDF
Ml master class
byQuantUniversity
 
PDF
Ml master class northeastern university
byQuantUniversity
 
PDF
AI Powerpoint Presentation Slides
bySlideTeam
 
PDF
AI PowerPoint Presentation Slides
bySlideTeam
 
PDF
Introduction to AI, ML, and Deep Learning - Workshop Presentation.pdf
byannsanson2025
 
PDF
Business Analytics PowerPoint Presentation Slides
bySlideTeam
 
PDF
Increasing Role Of Machine Intelligence Powerpoint Presentation Slides
bySlideTeam
 
PDF
Machine Intelligence PowerPoint Presentation Slides
bySlideTeam
 
PDF
Machine Intelligence Powerpoint Presentation Slides
bySlideTeam
 
PDF
Bi PowerPoint Presentation Slides
bySlideTeam
 
PDF
Principles of Artificial Intelligence & Machine Learning
byJerry Lu
 
PDF
Summit Australia 2019 - Supercharge PowerPlatform with AI - Dipankar Bhattach...
byAndrew Ly
 
PDF
Artificial Intelligence (2016) - AMP New Ventures
byAMP New Ventures
 
introductiontoartificialintelligencepowerpointpresentationslides-210722041732...
bySudhanshu Gupta
 
Introduction To Artificial Intelligence Powerpoint Presentation Slides
bySlideTeam
 
Artificial Intelligence And Machine Learning PowerPoint Presentation Slides C...
bySlideTeam
 
Back Propagation Neural Network In AI PowerPoint Presentation Slide Templates...
bySlideTeam
 
AI Vs ML Vs DL PowerPoint Presentation Slide Templates Complete Deck
bySlideTeam
 
What is artificial intelligence (IA) ?
byOussama Belakhdar
 
Introduction To Artificial Intelligence PowerPoint Presentation Slides
bySlideTeam
 
Ml master class
byQuantUniversity
 
Ml master class northeastern university
byQuantUniversity
 
AI Powerpoint Presentation Slides
bySlideTeam
 
AI PowerPoint Presentation Slides
bySlideTeam
 
Introduction to AI, ML, and Deep Learning - Workshop Presentation.pdf
byannsanson2025
 
Business Analytics PowerPoint Presentation Slides
bySlideTeam
 
Increasing Role Of Machine Intelligence Powerpoint Presentation Slides
bySlideTeam
 
Machine Intelligence PowerPoint Presentation Slides
bySlideTeam
 
Machine Intelligence Powerpoint Presentation Slides
bySlideTeam
 
Bi PowerPoint Presentation Slides
bySlideTeam
 
Principles of Artificial Intelligence & Machine Learning
byJerry Lu
 
Summit Australia 2019 - Supercharge PowerPlatform with AI - Dipankar Bhattach...
byAndrew Ly
 
Artificial Intelligence (2016) - AMP New Ventures
byAMP New Ventures
 

More from revathi148366

PPTX
ELECTRONIC DATA INTERCHANGE FOR ECOMMERCE.pptx
byrevathi148366
 
PPTX
BC(bLOCK CHAIN TECHNOLOGY)FOR STUDENTS.pptx
byrevathi148366
 
PPTX
Digital Token based Electronic payment system.pptx
byrevathi148366
 
PPTX
Evaluating the impact of Blockchain in Banking Sector.pptx
byrevathi148366
 
PPTX
ETL moel for data analytics modelling.pptx
byrevathi148366
 
PPTX
data compression for information technology.pptx
byrevathi148366
 
PPTX
search engine optimization for dmm1.pptx
byrevathi148366
 
PPTX
bank sector for blockchain application.pptx
byrevathi148366
 
PPTX
banking sector for blockchain presentation.pptx
byrevathi148366
 
PPTX
High frequency trading in deep learning.pptx
byrevathi148366
 
PPTX
Model training and parameter estimation techniques.pptx
byrevathi148366
 
PPTX
consumer oriented application in ecommerce.pptx
byrevathi148366
 
PPTX
digital marketing based epaymentsss.pptx
byrevathi148366
 
PPTX
digital marketing strategy 1st topic.pptx
byrevathi148366
 
PPTX
st of block chain technology 123556.pptx
byrevathi148366
 
PPTX
ecommerce architecure for www application.pptx
byrevathi148366
 
PPTX
architectural framework for ecommerce.pptx
byrevathi148366
 
PPT
Data Representationo for information.ppt
byrevathi148366
 
PPTX
working procedure of blockchaintech.pptx
byrevathi148366
 
PPTX
Deep Learning for financial technology.pptx
byrevathi148366
 
ELECTRONIC DATA INTERCHANGE FOR ECOMMERCE.pptx
byrevathi148366
 
BC(bLOCK CHAIN TECHNOLOGY)FOR STUDENTS.pptx
byrevathi148366
 
Digital Token based Electronic payment system.pptx
byrevathi148366
 
Evaluating the impact of Blockchain in Banking Sector.pptx
byrevathi148366
 
ETL moel for data analytics modelling.pptx
byrevathi148366
 
data compression for information technology.pptx
byrevathi148366
 
search engine optimization for dmm1.pptx
byrevathi148366
 
bank sector for blockchain application.pptx
byrevathi148366
 
banking sector for blockchain presentation.pptx
byrevathi148366
 
High frequency trading in deep learning.pptx
byrevathi148366
 
Model training and parameter estimation techniques.pptx
byrevathi148366
 
consumer oriented application in ecommerce.pptx
byrevathi148366
 
digital marketing based epaymentsss.pptx
byrevathi148366
 
digital marketing strategy 1st topic.pptx
byrevathi148366
 
st of block chain technology 123556.pptx
byrevathi148366
 
ecommerce architecure for www application.pptx
byrevathi148366
 
architectural framework for ecommerce.pptx
byrevathi148366
 
Data Representationo for information.ppt
byrevathi148366
 
working procedure of blockchaintech.pptx
byrevathi148366
 
Deep Learning for financial technology.pptx
byrevathi148366
 

Recently uploaded

PPTX
Ethics & Fairness Evaluation System with AI
byabhignareddykunduru
 
PDF
Effective Interest Region Estimation Model to Represent Corners for Image
bysipij
 
PPT
lecture 6,7,8- Circulatory physiology- Shock imp (2).ppt
byhambhatti123456
 
PDF
SaaS UX Design Best Practices & Ultimate Guide 2025.pdf
byDesign Studio UI UX
 
PPTX
National Art School Transcripts
by376x4gqj02
 
PPTX
Texture and patterns in fashion design.pptx
bySwami vivekanand university
 
PPTX
PPT expocision grupal de Ciencia y tecnologia proyectos
bycusialmendra
 
PDF
KONSTANTINOS V. PAPALAZAROU ⎯ A PRESENTATION OF AN INDUSTRY ICON
byKonstantinos V. Papalazarou
 
PPTX
Journey-into-Sound-An-Introduction.pptx.
bydhungelmanju796
 
PDF
Snapchat Ads Accounts_ Your Ultimate Resource.pdf
byBuy Verified OnlyFans Account
 
PPTX
A California Spanish Revival Home Defined by Floral Carved Antique Doors
byMogul Interior
 
PDF
Opium_Magazine_images_that_attack_the_eye.pdf
bybauermarija25
 
PPTX
NSX-T_and_Service_Interfaces_v2.1.1.pptx
bynandikumar4
 
PDF
Design House India is a renowned name for luxury residential and commercial i...
bydaytasksolutions
 
PDF
App Design Cost in 2025 A Complete Pricing Breakdown.pdf
byDesign Studio UI UX
 
PPTX
2022_gabe_hcx_icon_desing_ab-helper.pptx
bynandikumar4
 
PDF
Understanding Colour Psychology When Interior Decorating
byScott Anson Painters
 
PDF
Plasmodesmata.pdf........................
byi93297517
 
PDF
EITCI certification exam dumps and study materials
byCerts Area
 
PDF
Marriage Hall Banquet Marquee Front Design Services Pakistan.pdf
by3dfrontelevation.co Architect & Interior Designer & House Designer
 
Ethics & Fairness Evaluation System with AI
byabhignareddykunduru
 
Effective Interest Region Estimation Model to Represent Corners for Image
bysipij
 
lecture 6,7,8- Circulatory physiology- Shock imp (2).ppt
byhambhatti123456
 
SaaS UX Design Best Practices & Ultimate Guide 2025.pdf
byDesign Studio UI UX
 
National Art School Transcripts
by376x4gqj02
 
Texture and patterns in fashion design.pptx
bySwami vivekanand university
 
PPT expocision grupal de Ciencia y tecnologia proyectos
bycusialmendra
 
KONSTANTINOS V. PAPALAZAROU ⎯ A PRESENTATION OF AN INDUSTRY ICON
byKonstantinos V. Papalazarou
 
Journey-into-Sound-An-Introduction.pptx.
bydhungelmanju796
 
Snapchat Ads Accounts_ Your Ultimate Resource.pdf
byBuy Verified OnlyFans Account
 
A California Spanish Revival Home Defined by Floral Carved Antique Doors
byMogul Interior
 
Opium_Magazine_images_that_attack_the_eye.pdf
bybauermarija25
 
NSX-T_and_Service_Interfaces_v2.1.1.pptx
bynandikumar4
 
Design House India is a renowned name for luxury residential and commercial i...
bydaytasksolutions
 
App Design Cost in 2025 A Complete Pricing Breakdown.pdf
byDesign Studio UI UX
 
2022_gabe_hcx_icon_desing_ab-helper.pptx
bynandikumar4
 
Understanding Colour Psychology When Interior Decorating
byScott Anson Painters
 
Plasmodesmata.pdf........................
byi93297517
 
EITCI certification exam dumps and study materials
byCerts Area
 
Marriage Hall Banquet Marquee Front Design Services Pakistan.pdf
by3dfrontelevation.co Architect & Interior Designer & House Designer
 

artificialintelligencedata driven analytics23.pptx

  • 1.
  • 2.
    Introduction 2  What isAI?  Introduction to AI Levels?  Types of Artificial Intelligence  AI VS machine learning vs deep learning  Where is AI used?  AI use cases  Why is AI booming now?  AI trend in 2020 01
  • 3.
    Artificial Intelligence Transforming theNature of Work, Learning, and Learning to Work Artificial intelligence (AI) is a popular branch of computer science that concerns with building “intelligent” smart machines capable of performing intelligent tasks. With rapid advancements in deep learning and machine learning, the tech industry is transforming radically. Artificial Intelligence Machine Learning . Deep Learning 3
  • 4.
    Introduction to AILevels? Types of Artificial Intelligence Artificial Narrow Intelligence 4 Artificial Super Intelligence Artificial General Intelligence
  • 6.
    Artificial Intelligence 2020 2019 2018 65% . 35% . 2020 20192018 AI AI 2016 2017 2018 2019 2020 Artificial intelligence (AI) is a popular branch of computer science that concerns with building “intelligent” smart machines capable of performing intelligent tasks. 6 With rapid advancements in deep learning and machine learning, tech industry is transforming radically.
  • 7.
    Machine Learning Machine learningis a type of AI that enables machines to learn from data and deliver predictive models. The machine learning is not dependent on any explicit programming but the data fed into it. It is a complicated process. Based on the data you feed into machine learning algorithm and the training given to it, an output is delivered. A predictive algorithm will create a predictive model. This slide is 100% editable. Adapt it to your needs and capture your audience's attention. 7 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 8.
    Deep Learning Deep Learningis a subfield of machine learning that is concerned with algorithms inspired by the brain's structure & functions known as artificial neural networks A computer model can be taught using Deep Learning to run classification actions using pictures, texts or sounds as input 8
  • 9.
    Where is AI used? Real-time Operations Management Customer Services Risk Management& Analytics Customer Insight Pricing & Promotion Predictive Analytics Customer Experience Supply Chain Knowledge Creation Research & Development Fraud Detection Human Resources 9 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 10.
    AI Usecase inHealthCare AI and Robotics Training Research 10 This slide is 100% editable. Adapt it to your needs and capture your audience's attention. Early Detection Keeping Well End of Life Care Treatment Decision Making Diagnosis
  • 11.
    AI Use Casesin Human Resource Recruiting Dynamic Career Sites Smart Sourcing Onboarding Automated Messages Curated Videos Learning Curated Training Skill Development Engagement HR Chatbot Engagement Surveys 11 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 12.
    AI in Bankingfor Fraud Detection Cardholder Profiles Postings Payment System Nonmonetary System Analyst Workstation Payment and Non-Monetary Transactions 8 Authorization System Rules Definition Configuration Workstation Case Management Database Case Creation Module Neural Network Engine Scoring Engine Expert Authorization Response Module Expert Rules Base 6 Case Creation Rules Execute 1 Auth Request Expert Rules Execute 3 4 Auth Recommendation 22 Auth Request & Score 5 Transaction & Score 7 Case Information 12 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 13.
    AI in SupplyChain 16 Structured & Unstructured Information Regulator y Data B2B Transactio n Data Inventor y Data Multimedi a Data Sensor Data Logistics Data Trading Partner Data Social Medi a Data Digital Ecosystem Data Lake Pervasive Visibility Proactive Replenishment Predictive Maintenance Secure Device Maintenance Ecosystem Integration Unified Messaging Actionable Insights IIOT – Securely Provisionally People, System and Things Secure Access Via Identity Management for Transient Users Procurement Manufacturing Customers Service Logistics
  • 14.
    Ai Chatbots inHealthcare Search Engine Users learn to search for information Social Platforms Like Facebook connect users online 14 Smartphones Bring the internet online Healthbots Bring all of the above together for healthcare use cases Artificial Intelligence Self Learning machines becomes smarter the more they are used Messenger Apps Lets users chat anywhere, anytime App Eco-system Lets users download and use apps easily
  • 15.
    Why is AIbooming now? 8,097 7,540 7,336 4,680 4,201 3,714 3,655 3,564 3,169 2,000 0 4,000 6,000 8,000 Static Image Recognition, Classification and Tagging Algorithm Trading Strategy Performance Management Efficient, Scalable Processing of Patient Data Predictive Maintenance Object Identification, Text Query of Images detection, classification, tracking Automated Geophysical Content Distribution on Feature Detection Social Media Object detection & classification, avoidance, navigation 24% 22% 27% 37% 36% 51% 60% 60% 66% 71% 0 20 40 60 80 100 HR/Workforce Management Operational Environment Monitoring Through External Devices/Systems Logistics & Supply Chain Production Floor Systems Expediting Transactions Fleet Mobile External Communication Customer Relation/Interaction (i.e., chatbots) Security/Fraud Global AI Revenue Forecast by 2025, Ranked by Use Case in millions US Dollar Penetration of Artificial Intelligence Skills, by Country Organizations deploying AI, by Functional Areas Unite d States 100% 92% China 84% Indi a 54% Israe l 45% German y 15
  • 16.
    10 AI Trendin 2020 Robotic Process Automation AI will make Healthcare more Accurate Data Modeling will move to the Edge AI will Come for B2B Ai-powered Chatbots AI In Retail Aerospace and Flight Operations Controlled by AI AI Mediated Media and Entertainment Advanced Cybersecurity Automated Business Process “2020 AI Trends” 16
  • 17.
    Machine Learning 02  What isMachine Learning?  7 Steps of Machine learning  Machine Learning vs. Traditional Programming  How does machine learning work?  Machine learning Algorithms  Machine learning use cases  How to choose Machine Learning Algorithm  Why to use decision tree algorithm learning  Challenges and Limitations of Machine learning  Application of Machine learning  Why is machine learning important? 20
  • 18.
    Machine Learning Traditional Programming Data (Input ) Program Output Data (Input ) Output Machine Learning Program Machine Learningis the result of General AI that involves developing machines that can deliver results better than humans Input Data: Feed Learner Various Data Output Data: Present Rules “Learning” Machine Learning System 18
  • 19.
    7 Steps ofMachine Learning 01 02 03 04 05 06 07 Choosing a Model 19 Training Preparing that Data Evaluation Prediction Gathering Data Hyperparameter Tuning
  • 20.
    Machine Learning vs.Traditional Programming Prediction Result Compute r Dat a Handcrafte d Model Learning Mode l Prediction Result New Data Mode l Sample Data Expected Result Compute r Compute r Traditional Modelling Machine Learning 20
  • 21.
    How does MachineLearning Work? Collect Data Collect data from hospitals, health insurance companies, social service agencies, police and fire dept. Define Objectives Identify, the problem to be solved and create a clear objective. 21 Prepare Data Preparing data is a crucial step and involves building workflows to clean, match and blend the data. Select Algorithm Depend on the problem to be solved and the type of data an appropriate algorithm will be chosen. Train Model Data is fed as input and the algorithm configured with the required parameters. A percent of the data can be utilized to train the model. Integrate Model Publish the prepared experiment as a web service, so applications can use the model. Test Model The remaining data is utilized to test the model, for accuracy. Depending on the results, improvements, can be performed in the “Train model’ and/or “Select Algorithm” phases, iteratively.
  • 22.
    Machine Learning Algorithms KNN  Trees  Logistic Regression  Naïve-Bayes  SVM Reinforcement Categorical Machine Learning  Apriori  FP-Growth Associatio n Anlaysis Hidden Markov Model  SVD  PCA  K-means Clusterin g Unsupervised Regression  Linear  Polynomial Decision Tree Classificatio n Supervised Random Forest Continuous 22
  • 23.
    Machine Learning UseCases Energy Feedstock & Utilities  Power Usage Analytics  Seismic Data Processing  Your Text Here  Smart Grid Management  Energy Demand & Supply Optimization Financial Services  Risk Analytics & Regulation  Customer Segmentation  Your Text Here  Credit Worthiness Evaluation Travel & Hospitality  Aircraft Scheduling  Dynamic Pricing  Your Text Here  Traffic Patterns & Congestion Management Manufacturing  Predictive Maintenance or Condition Monitoring  Your Text Here  Demand Forecasting  Process Optimization  Telematics Retail  Predictive Inventory Planning  Recommendatio n Engines  Your Text Here  Customer ROI & Lifetime Value Healthcare & Life Sciences 23  Alerts & Diagnostics from Real-time Patient Data  Your Text Here  Predictive Health Management  Healthcare Provider Sentiment Analysis
  • 24.
    How to ChooseMachine Learning Algorithm What do you want to do with your Data? Algorithm Cheat Sheet Additional Requirements Accuracy Linearity Number of Parameters Training Time Number of Features How to Select Machine Learning Algorithms 24
  • 25.
    Why use DecisionTree Machine Learning Algorithm? To Classify Non-linear Relationship between Predictors & Response Linear Relationship between Predictors & Response Use c4.5 Implementatio n Use Standard Regression Tree Responsible Variable has only 2 Categories Use Standard Classification here Use c4.5 Implementatio n To Predict Responsible variable is Continuous Decision Trees 25 Response Variable has Multiple Categories
  • 26.
    Challenges and Limitationsof Machine learning Advantages Easily Identifies Trends and Patterns No Human Intervention needed Handling multi- dimensional & multi- variety Data Continuous Improvement Wide Applications Data Acquisitio n Time and Resource s High error- Susceptibilit y Interpretatio n Results Disadvantages 26
  • 27.
    Application of MachineLearning Automatic Language Translation Medical Diagnosis Stock Market Trading Online Fraud Detection Virtual Personal Assistant Email Spam and Malware Filtering Self Driving Cars Product Recommendations Traffic Prediction Speech Recognition Image Recognition 27
  • 28.
    Why is MachineLearning Important? Phase 1 : Learning Training Data Phase 2: Prediction  Precision/ recall  Over fitting  Test/cross Validation data, etc. Error Analysis  Normalization  Dimension Reduction  Image Processing, etc. Pre-Processing  Supervised  Unsupervised  Minimization, etc. Learning Predicted Data 28 Prediction New Data Model
  • 30.
    Deep Learning 03  Whatis Deep Learning?  Deep learning Process  Classification of Neural Networks  Types of Deep Learning Networks  Feed-forward neural networks  Recurrent neural networks (RNNs)  Convolutional neural networks (CNN)  Reinforcement Learning  Examples of deep learning applications  Why is Deep Learning Important?  Limitations of deep learning 32
  • 31.
    What is DeepLearning? Deep Learning is a subfield of machine learning that is concerned with algorithms inspired by the brain's structure and functions known as artificial neural networks. A computer model can be taught using Deep Learning to run classification actions using pictures, texts or sounds as input. Car Not Car Output Input Feature Extraction + Classification What is Deep Learning? 31
  • 32.
    Deep Learning Process Understand theProblem This slide is 100% editable. Adapt it to your needs and capture your audience's attention. 32 Identify Data This slide is 100% editable. Adapt it to your needs and capture your audience's attention. Select Deep Learning Algorithms This slide is 100% editable. Adapt it to your needs and capture your audience's attention. Training the Model This slide is 100% editable. Adapt it to your needs and capture your audience's attention. Test the Model This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 33.
    Classification of NeuralNetworks Input Layer Hidden Layer Output Layer x1 x2 xn v11 v12 vpn w11 w22 wmp ym y2 y1 V1n 1 2 p w1p 1 33 2 m
  • 34.
    Types of DeepLearning Networks Artificial Neural Networks Used for Regression & Classification Convolutional Neural Networks Used for Computer Vision Supervised Recurrent Neutral Networks Used for Time Series Analysis Self-Organizing Maps Used for Feature Detection Deep Boltzmann Machines Used for Recommendation Systems Unsupervised AutoEncoders Used for Recommendation Systems Deep Learning Models Supervised  Artificial Neural Networks (ANN)  Convolutional Neural Networks (CNN)  Recurrent Neural Networks (RNN) Unsupervised  Self Organizing Maps (SOM)  Boltzmann Machines (BM)  AutoEncoders (AE) 34
  • 35.
    Feed-forward Neural Networks InputLayer Hidden Layer Output Layer Variable- #1 35 Variable- #2 Variable- #3 Variable- # 4 Output
  • 36.
    Recurrent Neural Networks (RNNs) x1 x2 InputLayer 36 Recurrent Network Output Layer Hidden Layers
  • 37.
    Convolutional Neural Networks(CNN) Take Car = A1 37 Truck = B1 VAN = C1 Bicycle = D1 Rest all be Same
  • 38.
    Reinforcement Learning State, Reward Action AgentEnvironment Exploration Policy Neural Networks Filters Memor y Algorith m Reinforcement Learning uses rewards and punishment to train computing models to perform a sequence of selections. Here computing faces a game-like scenario where it employs trial and error to answer. Based on the action it performs, computing gets either rewards or penalties. Its goal is to maximize the rewards. 38
  • 39.
    Examples of DeepLearning Applications Image Recognitio n Natural Language Processing Portfolio Management & Prediction of Stock Price Movements Drug Discovery & Better Diagnostics of Diseases in Healthcare Speech Recognitio n Robots and Self - Driving Cars 39
  • 40.
    Why is DeepLearning Important? Performance Data Deep Learning Other Learning Algorithms 40
  • 41.
    Limitations of DeepLearning Limitations of Deep Learning Amount of Data Statistical Reasoning 41 Interpretability
  • 42.
    Difference between AI vsML vs DL 04  What is AI?  What is ML?  What is Deep Learning?  Machine Learning Process  Deep Learning Process  Difference between Machine Learning and Deep Learning  Which is better to start AI,ML or Deep learning 44
  • 43.
    Difference between AIvs ML vs DL Machine Learning Ability to learn without being explicitly programmed Deep Learnin g Learning based on deep neural network Artificial Intelligence Engineering of making intelligent machines and programs 43 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 44.
    What is AI? Artificial Intelligence Artificial Intelligence With rapidadvancements in deep learning and machine learning, the tech industry is transforming radically. (AI) is a popular branch of computer science that concerns with building “intelligent” smart machines capable of performing intelligent tasks. 44 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 45.
    What is ML? Learns Predicts Improves Machine Learning Ordinary System With AI Machine Learning isa type of AI that enables machines to learn from data and deliver predictive models. The machine learning is not dependent on any explicit programming but the data fed into it. It is a complicated process. Based on the data you feed into machine learning algorithm and the training given to it, an output is delivered. A predictive algorithm will create a predictive model. Introduction to Machine learning 45
  • 46.
    What is DeepLearning? Artificial intelligence (AI) is a popular branch of computer science that concerns with building “intelligent” smart machines capable of performing intelligent tasks. With rapid advancements in deep learning and machine learning, the tech industry is transforming radically. 46 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 47.
    Machine Learning Process Data Raw & TrainingData Modelling Candidate & Final Visualisation Predictions & Strategy Data Gathering Data Cleaning Selecting Right Algorithms Building Model & Finalising Data Transformation into Predictions 47 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 48.
    Deep Learning Process Testthe Model 48 Understand the Problem Identify Data Select Deep Learning Algorithm Training the Model
  • 49.
    Difference between MachineLearning and Deep Learning Inpu t Feature Extractio n Classifica -tion Outpu t Car Not Car Inpu t Feature Extraction + Classification Outpu t Car Not Car 49 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 50.
    Which is betterto start AI,ML or DL? Artificial Intelligence Any Technique which enables computers to mimic human behavior. 50 This slide is 100% editable. Adapt it to your needs and capture your audience's attention. Machine Learning Subset of AI Techniques which use Statistical Methods to Enable Machines to Improve with Experiences. Deep Learning Subset of ML which make the Computation of Multi- layer Neural Networks Feasible. 01 02 03
  • 51.
    Supervised Machine Learning 05  Types ofMachine Learning  What is Supervised Machine Learning?  How Supervised Learning Works  Types of Supervised Machine Learning Algorithms  Supervised vs. Unsupervised Machine learning techniques  Advantages of Supervised Learning:  Disadvantages of Supervised Learning 53
  • 52.
    Types of MachineLearning Inputs Training Inputs Makes Machine Learn Explicitly Data with Clearly defined Output is given Direct feedback is given Predicts outcome/future Resolves Classification and Regression Problems Supervised Learning Unsupervised Learning Machine Understands the data (Identifies Patterns/ Structures) Evaluation is Qualitative or Indirect Does not Predict/Find anything Specific Reinforcement Learning An approach to AI Reward Based Learning Learning form +ve & +ve Reinforcement Machine Learns how to act in a Certain Environment To Maximize Rewards Inputs Rewards Outputs 52 This slide is 100% editable. Adapt it to your needs and capture your audience's attention. Outputs Outputs
  • 53.
    What is SupervisedMachine Learning? Input Raw Data Output Processing Algorithm Training Data set Desired Output Supervisor Supervised Learning 53 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 54.
    How Supervised MachineLearning works Classification Sorting Items into Categories Regression Identifying Real Values (Dollars, Weight, etc.) Label “Group 1” Machine Step1 Provide the Machine Learning Algorithm Categorized or “labeled” Input and Output Data from to Learn Group 1 Group 2 Mach ine Step2 Feed the Machine New, Unlabeled Information to See if it Tags New Data Appropriately. If not, Continue Refining the Algorithm Types of Problems to which it’s Suited 54 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 55.
    Types of SupervisedMachine Learning Algorithms Classification  Fraud Detection  Email Spam Detection  Diagnostics  Image Classification. Regression  Risk Assessment  Score Prediction Supervised Learning 55 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 56.
    Supervised vs. UnsupervisedMachine Learning Techniques V S Supervised Learning Input & Output Data  Classification  Regression Predictions & Predictive Models Unsupervised Learning Input Data  Clustering  Association Patterns / Structure Discovery 56 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 57.
    It allows youto be very specific about the definition of the labels. In other words, you'll train the algorithm to differentiate different classes where you'll set a perfect decision boundary. You are ready to determine the amount of classes you would like to possess. 57 This slide is 100% editable. Adapt it to your needs and capture your audience's attention. The input file is extremely documented and is labeled. The results produced by the supervised method are more accurate and reliable as compared to the results produced by the unsupervised techniques of machine learning. this is often mainly because the input file within the supervised algorithm is documented and labeled. this is often a key difference between supervised and unsupervised learning. The answers within the analysis and therefore the output of your algorithm are likely to be known thanks to that each one the classes used are known. Advantages of Supervised Learning Advantages
  • 58.
    Disadvantages of SupervisedLearning Supervised learning are often a posh method as compared with the unsupervised method. The key reason is that you simply need to understand alright and label the inputs in supervised learning. 58 This slide is 100% editable. Adapt it to your needs and capture your audience's attention. It doesn’t happen in real time while the unsupervised learning is about the important time. this is often also a serious difference between supervised and unsupervised learning. Supervised machine learning uses of-line analysis. It is needed tons of computation time for training. If you've got a dynamic big and growing data, you're unsure of the labels to predefine the principles. this will be a true challenge.
  • 59.
    Unsupervised Machine Learning 06  What isUnsupervised Learning?  How Unsupervised Machine Learning works  Types of Unsupervised Learning  Disadvantages of Unsupervised Learning 61
  • 60.
    What is UnsupervisedLearning? Input Raw Data Output Algorithm Interpretation Processing  Unknown output  No Training Data Set Unsupervised Learning 60 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 61.
    How Unsupervised MachineLearning works Machine Step 1 Provide the machine learning algorithm uncategorized, unlabeled input data to see what patterns it finds Similar Group 1 Similar Group 2 Machine Step 2 Observe and learn from the patterns the machine identifies Types of Problems to Which it’s Suited Clustering Identifying similarities in groups For Example: Are there patterns in the data to indicate certain patients will respond better to this treatment than others? Anomaly Detection Identifying abnormalities in data For Example: Is a hacker intruding in our network? 61 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 62.
    Types of UnsupervisedLearning Dimensionality Reduction  Text Mining  Face Recognition  Big Data Visualization  Image Recognition Clustering  Biology  City Planning  Targeted Marketing Unsupervised Learning 62 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 63.
    Disadvantages of UnsupervisedLearning You cannot get very specific about the definition of the info sorting and therefore the output. This is often because the info utilized in unsupervised learning is labeled and not known. It's employment of the machine to label and group the data before determining the hidden patterns. 63 This slide is 100% editable. Adapt it to your needs and capture your audience's attention. Less accuracy of the results. This is often also because the input file isn't known and not labeled by people beforehand , which suggests that the machine will got to do that alone. The results of the analysis can't be ascertained. there's no prior knowledge within the unsupervised method of machine learning. Additionally, the numbers of classes also are not known. It results in the lack to determine the results generated by the analysis.
  • 64.
    Reinforcement learning 07  Whatis Unsupervised Learning?  How Unsupervised Machine Learning works  Types of Unsupervised Learning  Disadvantages of Unsupervised Learning 66
  • 65.
    What is Reinforcement Learning? Reinforced Response Input Input Response It’sa mang o Feedback Wrong! It’s an apple Learns Note d It’s an Apple 65 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 66.
    How Reinforcement Learning Works? InputRaw Data Rewar d State Selection of Algorith m Best Action Environment Agent Output Reinforcement Learning 66 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 67.
    Types of ReinforcementLearning Gaming Finance Sector 67 This slide is 100% editable. Adapt it to your needs and capture your audience's attention. Inventory Management Robot Navigation Manufacturing
  • 68.
    Disadvantage of ReinforcementLearning You cannot get very specific about the definition of the info sorting and therefore the output. this is often because the info utilized in unsupervised learning is labeled and not known. it's employment of the machine to label and group the data before determining the hidden patterns. Less accuracy of the results. this is often also because the input file isn't known and not labeled by people beforehand , which suggests that the machine will got to do that alone. The results of the analysis can't be ascertained. There's no prior knowledge within the unsupervised method of machine learning. Additionally, the numbers of classes also are not known. It results in the lack to determine the results generated by the analysis Reinforcement learning as a framework is wrong in many various ways, but it's precisely this quality that creates it useful. Too much reinforcement learning can cause an overload of states which may diminish the results. Reinforcement learning isn't preferable to use for solving simple problems Reinforcement learning needs tons of knowledge and tons of computation. it's data-hungry. that's why it works rather well in video games because one can play the sport again and again and again, so getting many data seems feasible. 68 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 69.
    Back Propagation Neural Networkin AI 08  Back Propagation Neural Network in AI  What is Artificial Neural Networks?  What is Backpropagation?  Why We Need Backpropagation?  What is a Feed Forward Network?  Types of Backpropagation Networks  Best practice Backpropagation 71
  • 70.
    Back Propagation NeuralNetwork in AI 1 2 i1 i2 h 2 w1 b1 b2 net out 70 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 71.
    What is ArtificialNeural Networks? Feed-Forward Network Output Input Layer Network Inputs Hidden Layer Back Propagation Output Layer 71 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 72.
    What is BackpropagationNeural Networking? x x x w w w w Difference in Desired Values 1 Input Layer Hidden Layer(s) 1 3 Output Layer 5 Backprop Output Layer 72 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 73.
    Why We NeedBackpropagation? 75 Most prominent advantages of Backpropagation Are: Backpropagation is fast, simple and straightforward to program. It has no parameters to tune aside from the numbers of input. It is a typical method that generally works well. It doesn't need any special mention of the features of the function to be learned. It is a versatile method because it doesn't require prior knowledge about the network.
  • 74.
    What is aFeed Forward Network? Input Layer Hidden Layer Output Layer 74 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 75.
    Types of BackpropagationNetworks Static Back-propagation Recurrent Backpropagation 75 This slide is 100% editable. Adapt it to your needs and capture your audience's attention. Static Back-propagation It is one quite backpropagation network which produces a mapping of a static input for static output. it's useful to unravel static classification issues like optical character recognition. Recurrent Backpropagation Recurrent backpropagation is fed forward until a hard and fast value is achieved. Then, the error is computed and propagated backward. The main difference between both of those methods is: that the mapping is rapid in static back-propagation while it's nonstatic in recurrent backpropagation
  • 76.
    Best Practice Backpropagation Aneural network is a group of connected it I/O units where each connection features a weight related to its computer programs. 76 Backpropagation is fast, simple and straightforward to program. A feedforward neural network is a man- made neural network. Backpropagation may be a short form for "backward propagation of errors." it's a typical method of coaching artificial neural networks.
  • 77.
    Expert System in ArtificialIntelligence 09  What is an Expert System?  Examples of Expert Systems  Characteristic of Expert System  Components of the expert system  Conventional System vs. Expert system  Human expert vs. expert system  Benefits of expert systems  Limitations of the expert system  Applications of expert systems 79
  • 78.
    Types of DeepLearning Networks The Expert System in AI are computer applications. Also, with the assistance of this development, we will solve complex problems. it's level of human intelligence and expertise Knowledge Base Inference Engine User Interface User (May not be an expert) Human Expert Knowledge Engineer 78
  • 79.
    Examples of Expert Systems TheHighest Level of Expertise The expert system offers the very best level of experience. It provides efficiency, accuracy and imaginative problem-solving. Right on Time Reaction An Expert System interacts during a very reasonable period of your time with the user. the entire time must be but the time taken by an expert to urge the foremost accurate solution for an equivalent problem Good Reliability The expert system must be reliable, and it must not make any an error. Flexible It is significant that it remains flexible because it the is possessed by an Expert system. Capable of Handling Challenging Decision & Problems An expert system is capable of handling challenging decision problems and delivering solutions. Effective Mechanism Expert System must have an efficient mechanism to administer the compilation of the prevailing knowledge in it. Expert System Non- expert User Knowledge from an expert User Interface Inference Engine Knowledge Base 79 Query Advice
  • 80.
    Characteristic of ExpertSystem High level Performance The system must be capable of responding at A level of competency adequate to or better than an expert system within the field. the standard of the recommendation given by the system should be during a high level integrity and that the performance ratio should be also very high Domain Specificity Expert systems are typically very domain specific. For ex., a diagnostic expert system for troubleshooting computers must actually perform all the required data manipulation as a person's expert would. The developer of such a system must limit his or her scope of the system to only what's needed to unravel the target problem. Special tools or programming languages are often needed to accomplish the precise objectives of the system Good Reliability The expert system must be as reliable as a person's expert Adequate Response Time The system should be designed in such how that it's ready to perform within alittle amount of your time , like or better than the time taken by a person's expert to succeed in at a choice point. An expert system that takes a year to succeed in a choice compared to a person's expert’s time of 1 hour wouldn't be useful Understandable The system should be understandable i.e. be ready to explain the steps of reasoning while executing. The expert system should have an evidence capability almost like the reasoning ability of human experts Use Symbolic Representations Expert systems use symbolic representations for knowledge (rules, networks or frames) and perform their inference through symbolic computations that closely resemble manipulations of tongue 80
  • 81.
    Components of theExpert System Explanation Inference Engine Knowledge Base Acquisition Facility User Interface Experts and Knowledge Engineers Users 81
  • 82.
    Conventional System vs.Expert System vs Knowledge domain break away the mechanism processing 01 The program could have made an error 02 Not necessarily need all the input/data 03 Changes within the rule are often made with ease 04 The system can work only with the rule as a tittle 05 Information and processing combined during a sequential file 01 The program isn't wrong 02 Need all the input file 03 Changes to the program are inconvenient 04 The system works if it's complete 05 82
  • 83.
    Human Expert vs.Expert System 85 Human Experts (Artificial ) Expert Systems Permanen t Perishabl e Easy to Transfer Difficult to Transfer Easy to Document Difficult to Document Affordable, costly to develop, but cheap to operate Expensive, especially top notch Add Your Text Here Add Your Text Here
  • 84.
    Benefits of Expert Systems 01 Easyto Develop and Modify the System 02 Fast Response 03 Low Accessibility Cost 04 Error Rate are Very Low 05 Humans Emotions are not Affected 84
  • 85.
    Limitations of theExpert System Don’t Have Decision Making Power Like Humans Expert System is not Widely used or Tested It cant Deal with the Mixed Knowledge There are Chances of Errors Its Difficult to Maintain Development Cost is High Not Able to Explain the Logic Behind the 85 This slide is 100% editable. Adapt it to your needs and capture your audience's attention.
  • 86.
    Applications of Expert Systems Medicaldomain (Diagnosis system, medical operations Process Control System Finance/Commerce (Stock market trading, airline scheduling cargo scheduling) Warehousin g Optimizatio n Knowledge domain (Finding out the faults in vehicles, computer) Repairin g Monitorin g system Design domain (Camera lens design,automobile design) Shipping 86
  • 87.
    Good ideas startwith brainstorming Great ideas start with coffee. Coffee Break 87
  • 88.
    Bar Chart 02 Product Thisgraph/chart is linked to excel, and changes automatically based on data. Just left click on it and select “Edit Data”. 01 Product This graph/chart is linked to excel, and changes automatically based on data. Just left click on it and select “Edit Data”. 100 90 80 70 60 50 40 30 20 10 0 88 Jan Feb Mar Apr May Jun Sales ( in USD millions) Year 2020 100% Product 01 Product 02
  • 89.
    Stacked Line WithMarkers 02 Product This graph/chart is linked to excel, and changes automatically based on data. Just left click on it and select “Edit Data”. 01 Product This graph/chart is linked to excel, and changes automatically based on data. Just left click on it and select “Edit Data”. 3628.4 89 3573.9 3484.0 3532.1 3740.3 3881.7 3528.4 3873.9 3584.0 3732.1 3640.3 3981.7 3400 3500 3600 3700 3800 3900 4000 4100 2015 2016 2017 2018 2019 2020 In Millions YEARS