HacktoberFestPune is a beginner-friendly, all-inclusive event that is absolutely free of cost. Certificates will be issued by DSC MESCOE and DSC PVGCOET for everyone who can complete 4 successful Pull Requests by 13th October 10 AM! An evening filled with speaker sessions, interactions with fellow developers, and mini-games, we think you'll have a great time with everyone!
2. Today’s Timeline
● Introduction to Hacktoberfest and Open Source
● Contributions Begin!
● Expert Talk — Mr. Akshay Kulkarni
● Expert Talk — Mr. Ashish Tetali
● Fun and Games
● Closing Ceremony
3. What isHacktoberfest?
●
● Hacktoberfest is a month-long celebration of open source
software presented by Digital Ocean, Intel, and Dev
Maintainers are invited to guide would-be contributors toward
issues that will help move the project forward, and
contributors get the opportunity to give back to both projects
they like and others they've just discovered
4. Why contribute to Open Source?
● Give back to the community
● Build new skills and gain experience
● Looks great on your resume and your GitHub profile
● Meet awesome people and network
— Join our Discord discord.gg/NvnYM9 to interact with other
participants.
● Turn your side project into a full-blown open source product
● It’s fun!
5. What is so special about Hacktoberfest?
● You get a free t-shirt for submitting 4 PRs between October 1st and
October 31st
● T-shirts will be awarded on a first-come, first-serve basis to the first
70,000 participants who successfully complete the Hacktoberfest
challenge
● Receive Certificates from us for submitting 4 PRs between October
11th and October 13th 10 AMIST.
● Please note, low quality/spammy contributions will not count
6. How do I find a way to contribute?
● You can go to our GitHub accounts for repositories you can
contribute to —
DSC MESCOE bit.ly/dsc-mescoe-github
AND
DSC PVGCOET bit.ly/dsc-pvgcoet-github
Note — Look for Repositories labelled “hacktoberfest” or Issues
labelled “hacktoberfest-accepted”
7. How do I find a way to contribute?
● We have various repositories in store for you from basic to
intermediate levels of development.
● Topics include:
1. JavaScript and Node
2. HTML/CSS
3. React
4. Flutter
5. ML/AI
6. Python
7. Data Structures and Algorithms
8. Rules for Hacktoberfest
● PRs must be made on the GitHub platform and merged. Only then, they
will count
● Only PRs count. Issues/commits do not
● All pull requests created between Oct. 1 and Oct. 31 will count, regardless
of when you register for Hacktoberfest
● Pull requests created before Oct. 1 but merged after do not count
9. Sounds awesome. I’m in. What’s next?
● Step 1
A. Register at bit.ly/Register-HacktoberFestPune if you haven’t
already. VVIMP to receive Certificates
B. Go to hacktoberfest.digitalocean.com and click on “Start Hacking”.
Log In with your GitHubAccount.
12. 12
ABOUT THE SPEAKER
Akshay Kulkarni
“Akshay Kulkarni is an Artificial Intelligence and Machine Learning Evangelist and also
he is a Google Developers Expert – Machine Learning, published author, of the books on
NLP and DL and regular speaker at major AI and Data Science conferences including
Strata, O’Reilly AI Conf, GIDS”
https://www.linkedin.com/in/akshay-kulkarni-1a562679/
28. • Churn prediction is one of the most
popular Big Data use cases in
business. It consists of detecting
customers who are likely to cancel a
subscription to a service
CHURN PREDICTION
29. • Customer segmentation is the practice
of dividing a customer base into groups of
individuals that are similar in specific ways
so companies can do marketing to each
group effectively and appropriately
CUSTOMER SEGMENTATION
30. RECOMMENDATION ENGINES
Business Problem
Personalized product recommendation based on user’s preference
Approach
Using Machine Learning algorithms ( hybrid recommendation
systems ) on customer purchase data along with third party
information
Using exhaustive recommendation strategies such as
Recommended for you
Frequently bought together
Recommendation on recent viewed items
Browsing history
Related to items you have viewed
Customer who have bought this also bought
There is a new version
Best selling in category
Results
• Nearly 35% of Amazon’s revenue is generated by its
recommendation engine
31. COMPLAINT E-MAIL CLASSIFICATION
Business Problem
Roughly 2000 emails to contact center per day
Requests for refunds, general feedback, website trouble
and more
Not able to focus on critical problems
Results
• Able to respond to urgent mails 4 times faster
• Saved money on headcount of contact center
32. PERSONALIZATION
Business Problem
Anticipate user’s preference and improve user
experience
Higher degree of personalization and enable higher level
of anticipatory design
Use Machine Learning to track user’s behavior and
predict next action based on purchase history,
preferences etc.
Preselect size, color, fixed price etc. in anticipation of
purchase
• User’s delight and simplify their lives
36. SMART SPEAKER READS EMOTIONS TO PLAY MUSIC
Business Problem
Moodbox is an emotionally intelligent speaker
system that uses artificial intelligence to learn to pick
the best music.
Results
Moodbox uses its artificial intelligence system to
work out what music users most want to listen to.
Users interact with the Moodbox by voice control,
and can keep a diary of their moods using the
speaker's smartphone app.
37. AI LAWYER SPEEDS UP LEGAL RESEARCH
Business Problem
Answer legal questions
Results
Ross Intelligence, which is built on IBM's super-
computer Watson, uses natural language processing to
answer legal questions
Uses natural language processing to answer legal
questions in a fraction of the time that it would take a
legal assistant.
Ross also monitors the law constantly to keep the user
updated about changes that might affect their case, so
they don't need to sift through the mass of legal news.
38. CHATBOT - BURGER KING
Business Problem
Find nearby outlet and take orders through chat
Results
• Not disclosed to public yet
39. VISUAL RETAIL ANALYTICS & TARGETING
Business Problem
More than 2000 stores in Great Britain itself
Use video analytics to monitor cleanliness, compliance and verify
campaign rollouts
Approach
Used Prism Skylab services to identify customer behavioral patterns
Create display level heatmaps to identify product interaction
Pathmaps to understand aggregate customer movement in key
areas of store
People counting to measure conversion
Results
Understand customer activity and foot traffic in real time
Resulted in 19% increase in sales
40. 40
• The aim is to successfully
classify the product
categories with the best
precision
• It is the biggest challenge to
tag a particular product or
SKU to its category
RETAIL PRODUCT CATEGORIZATION
41. REINFORCEMENT LEARNING : ANSWER FOR YOUR CUSTOMER JOURNEY PROBLEM
Historical states
Current states
Next possible states
Marketing
activity
1. Environment : Customers and
channels
2. Action : Marketing campaign through
any channel
3. Reward : Any metric that quantifies
conversion or towards conversion
4. State : Static and dynamic behavior
44. AI IN MARKETING
• Predicting Customer Behavior
• Churn Prediction
• Hyper Targeting
• Content Creation & Curation
• Predicting Lifetime Value (LTV)
• Buying propensity model
• Cross selling/Recommendation algorithms
• Website traffic forecasting - Forecasting
• Sentiment and trend analysis - NLP
• Market basket analysis - APIORI
• Chabot’s - NLP
• Marketing spend optimization / channel –
• Improve Customer Experience (CX) by
predicting user’s next actions and choices
• Using deep learning to identify behavioral
pattern based on camera footage inside and
outside store
• Optimize store operations and layout
45. AI IN RETAIL / ECOMMERCE
• Product Categorization
• Image recognition and understanding
(product catalog categorization)
• Fraud detection/ Anomaly detection
• Entity Resolution (do these three
accounts belong to the same person?)
• Pricing Optimization
• Demand forecasting (Supply and demand
analysis and forecast)
• Search Ranking
• Query expansion
• Image and text based search engine
• Categorizing emails - For example
complaint
• Credit risk
• Replenishment
46. AI IN MANUFACTURING
• Supply Chain Management
• Predictive Failure Analysis
• Process Optimization
• Predictive Maintenance
• Image recognition, this can be utilized to
identify (classify) damaged products
• Logistics
47. AI IN AUTOMOTIVE
• Effective incorporation of analysis –
Product design
• Procurement
• Workforce analytics
• Parts pricing opt
• Operation planning
• Failure patterns to establish relationship
between failures & causes of failure
• Enhancing overall in-vehicle user
experience through personalization
• Prediction of battery life for electric
vehicles
48. AI IN HEATHCARE
• Patient life cycle in hospital
• Understanding Medical Data
• Outbreak Prediction
• Staff optimization – patients
expected
• Predicting diabetes
• Identifying fraud – insurance
providers
• Medical image analysis
55. 55
Why NLP is hard?
• Semantics “Gabe invited me to his medical school ball”. What is “ball” in this context?
• Morphology “dog not like eat vegetable”
• the dog did not like to eat vegetables
• the dogs do not like to eat vegetables
• the dogs did not like to eat vegetables
• dogs do not like to eat vegetables
• Ambiguity of intent: “I just got back from New York”. What do they want?
• Situational ambiguity: “Elaina was found by the river head”. Could be by the head of the river (place) or the executive of the river (person)
• Unable to deduce meaning of unknown words from context like humans can
• Disambiguation – “jaguar” can refer to a car or to an animal