Words are no longer sufficient in delivering the search results users are looking for, particularly in relation to image search. Text and languages pose many challenges in describing visual details and providing the necessary context for optimal results. Machine Learning technology opens a new world of search innovation that has yet to be applied by businesses.
In this session, Mike Ranzinger of Shutterstock will share a technical presentation detailing his research on composition aware search. He will also demonstrate how the research led to the launch of AI technology allowing users to more precisely find the image they need within Shutterstock’s collection of more than 150 million images. While the company released a number of AI search enabled tools in 2016, this new technology allows users to search for items in an image and specify where they should be located within the image. The research identifies the networks that localize and describe regions of an image as well as the relationships between things. The goal of this research was to improve the future of search using visual data, contextual search functions, and AI. A combination of multiple machine learning technologies led to this breakthrough.
Embed, Encode, Attend, Predict – applying the 4 step NLP recipe for text clas...Sujit Pal
Slides for talk at PyData Seattle 2017 about Matthew Honnibal's 4-step recipe for Deep Learning NLP pipelines. Description of the stages in pipeline as well as 3 examples of document classification, document similarity and sentence similarity. Examples include Keras custom layers for different types of attention.
With the explosive growth of online information, recommender system has been an effective tool to overcome information overload and promote sales. In recent years, deep learning's revolutionary advances in speech recognition, image analysis and natural language processing have gained significant attention. Meanwhile, recent studies also demonstrate its efficacy in coping with information retrieval and recommendation tasks. Applying deep learning techniques into recommender system has been gaining momentum due to its state-of-the-art performance. In this talk, I will present recent development of deep learning based recommender models and highlight some future challenges and open issues of this research field.
Embed, Encode, Attend, Predict – applying the 4 step NLP recipe for text clas...Sujit Pal
Slides for talk at PyData Seattle 2017 about Matthew Honnibal's 4-step recipe for Deep Learning NLP pipelines. Description of the stages in pipeline as well as 3 examples of document classification, document similarity and sentence similarity. Examples include Keras custom layers for different types of attention.
With the explosive growth of online information, recommender system has been an effective tool to overcome information overload and promote sales. In recent years, deep learning's revolutionary advances in speech recognition, image analysis and natural language processing have gained significant attention. Meanwhile, recent studies also demonstrate its efficacy in coping with information retrieval and recommendation tasks. Applying deep learning techniques into recommender system has been gaining momentum due to its state-of-the-art performance. In this talk, I will present recent development of deep learning based recommender models and highlight some future challenges and open issues of this research field.
Deep Learning in Recommender Systems - RecSys Summer School 2017Balázs Hidasi
This is the presentation accompanying my tutorial about deep learning methods in the recommender systems domain. The tutorial consists of a brief general overview of deep learning and the introduction of the four most prominent research direction of DL in recsys as of 2017. Presented during RecSys Summer School 2017 in Bolzano, Italy.
Deep Learning in Robotics
- There are two major branches in applying deep learning techniques in robotics.
- One is to combine DL with Q learning algorithms. For example, awesome work on playing Atari games done by deep mind is a representative study. While this approach can effectively handle several problems that can hardly be solved via traditional methods, these methods are not appropriate for real manipulators as it often requires an enormous number of training data.
- The other branch of work uses a concept of guided policy search. It combines trajectory optimization methods with supervised learning algorithm like CNNs to come up with a robust 'policy' function that can actually be used in real robots, e.g., Baxter of PR2.
Artificial Intelligence, Machine Learning and Deep LearningSujit Pal
Slides for talk Abhishek Sharma and I gave at the Gennovation tech talks (https://gennovationtalks.com/) at Genesis. The talk was part of outreach for the Deep Learning Enthusiasts meetup group at San Francisco. My part of the talk is covered from slides 19-34.
This is a slide deck from a presentation, that my colleague Shirin Glander (https://www.slideshare.net/ShirinGlander/) and I did together. As we created our respective parts of the presentation on our own, it is quite easy to figure out who did which part of the presentation as the two slide decks look quite different ... :)
For the sake of simplicity and completeness, I just copied the two slide decks together. As I did the "surrounding" part, I added Shirin's part at the place when she took over and then added my concluding slides at the end. Well, I'm sure, you will figure it out easily ... ;)
The presentation was intended to be an introduction to deep learning (DL) for people who are new to the topic. It starts with some DL success stories as motivation. Then a quick classification and a bit of history follows before the "how" part starts.
The first part of the "how" is some theory of DL, to demystify the topic and explain and connect some of the most important terms on the one hand, but also to give an idea of the broadness of the topic on the other hand.
After that the second part dives deeper into the question how to actually implement DL networks. This part starts with coding it all on your own and then moves on to less coding step by step, depending on where you want to start.
The presentation ends with some pitfalls and challenges that you should have in mind if you want to dive deeper into DL - plus the invitation to become part of it.
As always the voice track of the presentation is missing. I hope that the slides are of some use for you, though.
Part 2 of the Deep Learning Fundamentals Series, this session discusses Tuning Training (including hyperparameters, overfitting/underfitting), Training Algorithms (including different learning rates, backpropagation), Optimization (including stochastic gradient descent, momentum, Nesterov Accelerated Gradient, RMSprop, Adaptive algorithms - Adam, Adadelta, etc.), and a primer on Convolutional Neural Networks. The demos included in these slides are running on Keras with TensorFlow backend on Databricks.
Image generation. Gaussian models for human faces, limits and relations with linear neural networks. Generative adversarial networks (GANs), generators, discrinators, adversarial loss and two player games. Convolutional GAN and image arithmetic. Super-resolution. Nearest-neighbor, bilinear and bicubic interpolation. Image sharpening. Linear inverse problems, Tikhonov and Total-Variation regularization. Super-Resolution CNN, VDSR, Fast SRCNN, SRGAN, perceptual, adversarial and content losses. Style transfer: Gatys model, content loss and style loss.
This presentation introduces Deep Learning (DL) concepts, such as neural neworks, backprop, activation functions, and Convolutional Neural Networks, followed by an Angular application that uses TypeScript in order to replicate the Tensorflow playground.
In this talk we explore how to build Machine Learning Systems that can that can learn "continuously" from their mistakes (feedback loop) and adapt to an evolving data distribution.
The youtube link to video of the talk is here:
https://www.youtube.com/watch?v=VtBvmrmMJaI
An introduction to Machine Learning (and a little bit of Deep Learning)Thomas da Silva Paula
25-min talk about Machine Learning and a little bit of Deep Learning. Starts with some basic definitions (Supervised and Unsupervised Learning). Then, neural networks basic functionality is explained, ending up in Deep Learning and Convolutional Neural Networks.
Machine Learning Meetup that happened in Porto Alegre, Brazil.
Picked-up lists of GAN variants which provided insights to the community. (GANs-Improved GANs-DCGAN-Unrolled GAN-InfoGAN-f-GAN-EBGAN-WGAN)
After short introduction to GANs, we look through the remaining difficulties of standard GANs and their temporary solutions (Improved GANs). By following the slides, we can see the other solutions which tried to resolve the problems in various ways, e.g. careful architecture selection (DCGAN), slight change in update (Unrolled GAN), additional constraint (InfoGAN), generalization of the loss function using various divergence (f-GAN), providing new framework of energy based model (EBGAN), another step of generalization of the loss function (WGAN).
Deep Learning in Recommender Systems - RecSys Summer School 2017Balázs Hidasi
This is the presentation accompanying my tutorial about deep learning methods in the recommender systems domain. The tutorial consists of a brief general overview of deep learning and the introduction of the four most prominent research direction of DL in recsys as of 2017. Presented during RecSys Summer School 2017 in Bolzano, Italy.
Deep Learning in Robotics
- There are two major branches in applying deep learning techniques in robotics.
- One is to combine DL with Q learning algorithms. For example, awesome work on playing Atari games done by deep mind is a representative study. While this approach can effectively handle several problems that can hardly be solved via traditional methods, these methods are not appropriate for real manipulators as it often requires an enormous number of training data.
- The other branch of work uses a concept of guided policy search. It combines trajectory optimization methods with supervised learning algorithm like CNNs to come up with a robust 'policy' function that can actually be used in real robots, e.g., Baxter of PR2.
Artificial Intelligence, Machine Learning and Deep LearningSujit Pal
Slides for talk Abhishek Sharma and I gave at the Gennovation tech talks (https://gennovationtalks.com/) at Genesis. The talk was part of outreach for the Deep Learning Enthusiasts meetup group at San Francisco. My part of the talk is covered from slides 19-34.
This is a slide deck from a presentation, that my colleague Shirin Glander (https://www.slideshare.net/ShirinGlander/) and I did together. As we created our respective parts of the presentation on our own, it is quite easy to figure out who did which part of the presentation as the two slide decks look quite different ... :)
For the sake of simplicity and completeness, I just copied the two slide decks together. As I did the "surrounding" part, I added Shirin's part at the place when she took over and then added my concluding slides at the end. Well, I'm sure, you will figure it out easily ... ;)
The presentation was intended to be an introduction to deep learning (DL) for people who are new to the topic. It starts with some DL success stories as motivation. Then a quick classification and a bit of history follows before the "how" part starts.
The first part of the "how" is some theory of DL, to demystify the topic and explain and connect some of the most important terms on the one hand, but also to give an idea of the broadness of the topic on the other hand.
After that the second part dives deeper into the question how to actually implement DL networks. This part starts with coding it all on your own and then moves on to less coding step by step, depending on where you want to start.
The presentation ends with some pitfalls and challenges that you should have in mind if you want to dive deeper into DL - plus the invitation to become part of it.
As always the voice track of the presentation is missing. I hope that the slides are of some use for you, though.
Part 2 of the Deep Learning Fundamentals Series, this session discusses Tuning Training (including hyperparameters, overfitting/underfitting), Training Algorithms (including different learning rates, backpropagation), Optimization (including stochastic gradient descent, momentum, Nesterov Accelerated Gradient, RMSprop, Adaptive algorithms - Adam, Adadelta, etc.), and a primer on Convolutional Neural Networks. The demos included in these slides are running on Keras with TensorFlow backend on Databricks.
Image generation. Gaussian models for human faces, limits and relations with linear neural networks. Generative adversarial networks (GANs), generators, discrinators, adversarial loss and two player games. Convolutional GAN and image arithmetic. Super-resolution. Nearest-neighbor, bilinear and bicubic interpolation. Image sharpening. Linear inverse problems, Tikhonov and Total-Variation regularization. Super-Resolution CNN, VDSR, Fast SRCNN, SRGAN, perceptual, adversarial and content losses. Style transfer: Gatys model, content loss and style loss.
This presentation introduces Deep Learning (DL) concepts, such as neural neworks, backprop, activation functions, and Convolutional Neural Networks, followed by an Angular application that uses TypeScript in order to replicate the Tensorflow playground.
In this talk we explore how to build Machine Learning Systems that can that can learn "continuously" from their mistakes (feedback loop) and adapt to an evolving data distribution.
The youtube link to video of the talk is here:
https://www.youtube.com/watch?v=VtBvmrmMJaI
An introduction to Machine Learning (and a little bit of Deep Learning)Thomas da Silva Paula
25-min talk about Machine Learning and a little bit of Deep Learning. Starts with some basic definitions (Supervised and Unsupervised Learning). Then, neural networks basic functionality is explained, ending up in Deep Learning and Convolutional Neural Networks.
Machine Learning Meetup that happened in Porto Alegre, Brazil.
Picked-up lists of GAN variants which provided insights to the community. (GANs-Improved GANs-DCGAN-Unrolled GAN-InfoGAN-f-GAN-EBGAN-WGAN)
After short introduction to GANs, we look through the remaining difficulties of standard GANs and their temporary solutions (Improved GANs). By following the slides, we can see the other solutions which tried to resolve the problems in various ways, e.g. careful architecture selection (DCGAN), slight change in update (Unrolled GAN), additional constraint (InfoGAN), generalization of the loss function using various divergence (f-GAN), providing new framework of energy based model (EBGAN), another step of generalization of the loss function (WGAN).
NIT Silchar ML Hackathon 2019 Session on Computer Vision with Deep Learning.
Targeted Audience: Pre-requisite: Basic knowledge on Machine Learning and Deep Learning
Learning a Joint Embedding Representation for Image Search using Self-supervi...Sujit Pal
Image search interfaces either prompt the searcher to provide a search image (image-to-image search) or a text description of the image (text-to-image search). Image to Image search is generally implemented as a nearest neighbor search in a dense image embedding space, where the embedding is derived from Neural Networks pre-trained on a large image corpus such as ImageNet. Text to image search can be implemented via traditional (TF/IDF or BM25 based) text search against image captions or image tags.
In this presentation, we describe how we fine-tuned the OpenAI CLIP model (available from Hugging Face) to learn a joint image/text embedding representation from naturally occurring image-caption pairs in literature, using contrastive learning. We then show this model in action against a dataset of medical image-caption pairs, using the Vespa search engine to support text based (BM25), vector based (ANN) and hybrid text-to-image and image-to-image search.
DELAB - sequence generation seminar
Title
[Paper Review] Knowing when to look: Adaptive Attention via A Visual Sentinel for Image Captioning
Table of contents
1. Image Captioning
2. Knowing When to Look: Adaptive Attention via A Visual
Sentinel for Image Captioning
3. Model Architecture
1) Encoder-Decoder for Image Captioning
2) Spatial Attention Model
3) Adaptive Attention Model
4. Results
5. Adaptive Attention Analysis
Dr. Erin LeDell, Machine Learning Scientist, H2O.ai at MLconf SEA - 5/20/16MLconf
Multi-algorithm Ensemble Learning at Scale: Software, Hardware and Algorithmic Approaches: Multi-algorithm ensemble machine learning methods are often used when the true prediction function is not easily approximated by a single algorithm. The Super Learner algorithm, also known as stacking, combines multiple, typically diverse, base learning algorithms into a single, powerful prediction function through a secondary learning process called metalearning. Although ensemble methods offer superior performance over their singleton counterparts, there is an implicit computational cost to ensembles, as it requires training and cross-validating multiple base learning algorithms.
We will demonstrate a variety of software- and hardware-based approaches that lead to more scalable ensemble learning software, including a highly scalable implementation of stacking called “H2O Ensemble”, built on top of the open source, distributed machine learning platform, H2O. H2O Ensemble scales across multi-node clusters and allows the user to create ensembles of deep neural networks, Gradient Boosting Machines, Random Forest, and others. As for algorithm-based approaches, we will present two algorithmic modifications to the original stacking algorithm that further reduce computation time — Subsemble algorithm and the Online Super Learner algorithm. This talk will also include benchmarks of the implementations of these new stacking variants.
Automatic Attendace using convolutional neural network Face Recognitionvatsal199567
Automatic Attendance System will recognize the face of the student through the camera in the class and mark the attendance. It was built in Python with Machine Learning.
Erin LeDell, Machine Learning Scientist, H2O.ai at MLconf ATL 2016MLconf
Multi-algorithm Ensemble Learning at Scale: Software, Hardware and Algorithmic Approaches: Multi-algorithm ensemble machine learning methods are often used when the true prediction function is not easily approximated by a single algorithm. The Super Learner algorithm, also known as stacking, combines multiple, typically diverse, base learning algorithms into a single, powerful prediction function through a secondary learning process called metalearning. Although ensemble methods offer superior performance over their singleton counterparts, there is an implicit computational cost to ensembles, as it requires training and cross-validating multiple base learning algorithms.
We will demonstrate a variety of software- and hardware-based approaches that lead to more scalable ensemble learning software, including a highly scalable implementation of stacking called “H2O Ensemble”, built on top of the open source, distributed machine learning platform, H2O. H2O Ensemble scales across multi-node clusters and allows the user to create ensembles of deep neural networks, Gradient Boosting Machines, Random Forest, and others. As for algorithm-based approaches, we will present two algorithmic modifications to the original stacking algorithm that further reduce computation time — Subsemble algorithm and the Online Super Learner algorithm. This talk will also include benchmarks of the implementations of these new stacking variants.
Render 2D graphics in cross-platform mobile apps with HTML5, JavaScript, jQuery and PhoneGap. Learn how to draw in a way that will work across a variety of devices.
Similar to The Search for a New Visual Search Beyond Language - StampedeCon AI Summit 2017 (20)
Why Should We Trust You-Interpretability of Deep Neural Networks - StampedeCo...StampedeCon
Despite widespread adoption and success most machine learning models remain black boxes. Many times users and practitioners are asked to implicitly trust the results. However understanding the reasons behind predictions is critical in assessing trust, which is fundamental if one is asked to take action based on such models, or even to compare two similar models. In this talk I will (1.) formulate the notion of interpretability of models, (2.) provide a review of various attempts and research initiatives to solve this very important problem and (3.) demonstrate real industry use-cases and results focusing primarily on Deep Neural Networks.
Predicting Outcomes When Your Outcomes are Graphs - StampedeCon AI Summit 2017StampedeCon
In many modern applications data are collected in unusual form. Connectome or brain imaging data are graphs. Wearable devices measuring activity are functions over time. In many cases these objects are collected for each individual or transaction leaving the statistician with the challenge of analyzing populations of data not in classical numeric and categorical formats in big spreadsheets. In this talk I introduce object oriented data analysis with an application we recently developed for regression analysis. This talk will be aimed at the general data scientist and emphasis on the concepts and not mathematical detail. The take home message is how can we use covariates (i.e., meta-data) to predict what the structure of a brain image graph will be.
Novel Semi-supervised Probabilistic ML Approach to SNP Variant Calling - Stam...StampedeCon
This talk aims to dive into technical details in machine learning model development, implementation and values it bring to Monsanto breeding pipeline. We genotype over 100 million seeds a year in order to save field resources and product development cycle time. Automation and high throughput production from the lab becomes key to R&D success. In house predictive model development incorporated random forest ensemble based approach with additional features derived from gaussian mixture model. The results show over 95% accuracy with less than 1% false positives/negatives. Model is highly generalizable with over 10 million data points being trained and tested on. The model also offers probabilistic approach to present genotypes in a more meaningful way and help enhanced downstream genomics analyses. The talk targets audience who are in breeding, genetics, molecular biology, and data scientists who are interested in practical applications.
How to Talk about AI to Non-analaysts - Stampedecon AI Summit 2017StampedeCon
While artificial intelligence for self-driving cars and virtual assistants gets a lot of the notion of communicating the needs, effectiveness and measurements is complicated when speaking “geek”! The work of an analyst, however, does not just involve conducting data analysis within but communicating, championing and speaking simply when talking to the organization, clients and management.
Getting Started with Keras and TensorFlow - StampedeCon AI Summit 2017StampedeCon
This technical session provides a hands-on introduction to TensorFlow using Keras in the Python programming language. TensorFlow is Google’s scalable, distributed, GPU-powered compute graph engine that machine learning practitioners used for deep learning. Keras provides a Python-based API that makes it easy to create well-known types of neural networks in TensorFlow. Deep learning is a group of exciting new technologies for neural networks. Through a combination of advanced training techniques and neural network architectural components, it is now possible to train neural networks of much greater complexity. Deep learning allows a model to learn hierarchies of information in a way that is similar to the function of the human brain.
Foundations of Machine Learning - StampedeCon AI Summit 2017StampedeCon
This presentation will cover all aspects of modeling, from preparing data, training and evaluating the results. There will be descriptions of the mainline ML methods including, neural nets, SVM, boosting, bagging, trees, forests, and deep learning. common problems of overfitting and dimensionality will be covered with discussion of modeling best practices. Other topics will include field standardization, encoding categorical variables, feature creation and selection. It will be a soup-to-nuts overview of all the necessary procedures for building state-of-the art predictive models.
Don't Start from Scratch: Transfer Learning for Novel Computer Vision Problem...StampedeCon
In this session, we’ll discuss approaches for applying convolutional neural networks to novel computer vision problems, even without having millions of images of your own. Pretrained models and generic image data sets from Google, Kaggle, universities, and other places can be leveraged and adapted to solve industry and business specific problems. We’ll discuss the approaches of transfer learning and fine tuning to help anyone get started on using deep learning to get cutting edge results on their computer vision problems.
Bringing the Whole Elephant Into View Can Cognitive Systems Bring Real Soluti...StampedeCon
Like the story of the six blind men trying to explain the nature of an elephant, current research in cognitive computational systems attempts to identify the nature of an illness, human behavior, or socio-economical phenomenon, from their own perspective.
At present, there is no agreed upon definition for cognitive systems. One large communication corporation defines cognitive systems as a category of technology that uses artificial intelligence, machine learning and reasoning, to enable people and machines to interact more naturally. It also extends and magnifies human expertise and cognition to enable accurate decisions on time. Two of the most famous risk and financial advisory firms agree with that interpretation. A different large corporation, however, considers “cognitive systems” as merely marketing jargon.
If cognitive systems are going to help us solve challenging problems in medicine, economics, or other fields, three aspects must be considered in order to reveal the “true nature of the elephant”.
§ All facets of the problem must be addressed, like the main parts of the elephant had to be touched by the men.
§ These facets must be properly assembled, like the men needed to join hands around the elephant in order to understand what it was.
§ This assembly must be completed within sufficient time to anticipate future decisions. Just like the men needed to know what an elephant is before the next one charges them.
This talk will explain how agnostic (unsupervised, blinded) machine learning findings can be assembled by multiobjective and multimodal optimization research techniques would be utilized to uncover a multifaceted view of the “elephant”, in this case the human being (e.g., genomic variants, personality traits, brain images). It will also give real-world examples of how this knowledge will “extend the human capabilities” by achieving an integrative assessment of the whole person in relation to their risk, which will allow professionals to generate accurate person-centered policies: from personalized diagnoses, business opportunities, or the prevention of outbreaks.
Automated AI The Next Frontier in Analytics - StampedeCon AI Summit 2017StampedeCon
This talk will walk through the important building blocks of Automated AI. Rajiv will highlight the current gaps in the analytics organizations, how to close those gaps using automated AI. Some of the issues discussed around automated AI are the accuracy of models, tradeoffs around control when using automation, interpretability of models, and integration with other tools. These issues will be highlighted with examples of automated analytics in different industries. The talk will end with some examples of how automated AI in the hands of data scientists and business analysts is transforming analytic teams and organizations.
AI in the Enterprise: Past, Present & Future - StampedeCon AI Summit 2017StampedeCon
Artificial Intelligence has entered a renaissance thanks to rapid progress in domains as diverse as self-driving cars, intelligent assistants, and game play. Underlying this progress is Deep Learning – driven by significant improvements in Graphic Processing Units and computational models inspired by the human brain that excel at capturing structures hidden in massive complex datasets. These techniques have been pioneered at research universities and digital giants but mainstream enterprises are starting to apply them as open source tools and improved hardware become available. Learn how AI is impacting analytics today and in the future.
Learn how AI is affecting the enterprise including applications like fraud detection, mobile personalization, predicting failures for IoT and text analysis to improve call center interactions. We look at how practical examples of assessing the opportunity for AI, phased adoption, and lessons going from research, to prototype, to scaled production deployment.
A Different Data Science Approach - StampedeCon AI Summit 2017StampedeCon
This session will focus on how to execute Data Science caliber efforts by creating teams with the attributes of Data Science to deliver meaningful results. As Data Scientists are harder to find and keep, this session should appeal to anyone who is either seeking an alternative approach to executing Data Science delivery or augmenting their current Data Science model with additional options.
Graph in Customer 360 - StampedeCon Big Data Conference 2017StampedeCon
Enterprises typically have many data silos of partial customer data and a common theme in big data projects to use big data tools and pipelines to unify all siloed customer data into a single, queryable, platform for improving all future customer interactions. This data often comes from billing, website traffic, logistics, and marketing; all in different formats with different properties. Graph provides a way to unify all of the data into a single place for use in tracking the flow of a user through the various silos. Graph can also be used for visualizations and analytics that are difficult in other systems.
In this talk we will explore the ways in which Graph can be leveraged in a customer 360 use case. What it can add to a more conventional system and what the approach to developing a graph based Customer 360 system should be.
End-to-end Big Data Projects with Python - StampedeCon Big Data Conference 2017StampedeCon
This talk will go over how to build an end-to-end data processing system in Python, from data ingest, to data analytics, to machine learning, to user presentation. Developments in old and new tools have made this particularly possible today. The talk in particular will talk about Airflow for process workflows, PySpark for data processing, Python data science libraries for machine learning and advanced analytics, and building agile microservices in Python.
System architects, software engineers, data scientists, and business leaders can all benefit from attending the talk. They should learn how to build more agile data processing systems and take away some ideas on how their data systems could be simpler and more powerful.
Doing Big Data Using Amazon's Analogs - StampedeCon Big Data Conference 2017StampedeCon
Big Data doesn’t have to just mean Hadoop any more. Big Data can be done in the cloud, using tools developed by the Cloud providers. This session will cover using Amazon AWS services to implement a Big Data application. We will compare and contrast different services from Amazon with the Hadoop equivalents.
Enabling New Business Capabilities with Cloud-based Streaming Data Architectu...StampedeCon
Using big data isn’t about doing the same things we’ve always done just with different technologies. The technology advances that we’ve chosen to label as big data create the opportunity for wholly new kinds of solutions. Two of the key advances that are enabling new business capabilities are cloud-based data management platforms and streaming data processing and analytics.
In this session, Paul Boal will drill into the cloud-based streaming data architecture that has made possible EVŌ, a new breakthrough health and wellness platform. EVŌ uses a game-changing approach that leverages over 60 billion data points and a predictive analytics engine to intervene BEFORE someone becomes critically ill. All of this is possible by leveraging data from smartphones and wearable fitness devices along with advanced analytics which then help users develop and sustain positive behaviors. Attendees will learn how to create a cloud- based architecture that can receive data, apply multiple layers of dynamic business rules, and drive alerts and decisions through real-time stream processing using technologies including web services, Amazon DynamoDB and Kinesis, Drools, and Apache Spark.
Big Data Meets IoT: Lessons From the Cloud on Polling, Collecting, and Analyz...StampedeCon
The collection and use of Big Data has become an important part of modern business practice. The Internet of Things (IoT) movement promises to provide new opportunities for businesses interested in the intersection of people and technology. It is also wrought with pitfalls for practitioners and researchers who struggle to make sense of an increasing cacophony of signals. How should they poll and collect data from millions of signals in a way that is manageable, scalable, and statistically valid? How should they analyze and predict using these data? This presentation will discuss these challenges with applied examples from monitoring and managing one of the world’s largest computers.
Innovation in the Data Warehouse - StampedeCon 2016StampedeCon
Enterprise Holding’s first started with Hadoop as a POC in 2013. Today, we have clusters on premises and in the cloud. This talk will explore our experience with Big Data and outline three common big data architectures (batch, lambda, and kappa). Then, we’ll dive into the decision points to necessary for your own cluster, for example: cloud vs on premises, physical vs virtual, workload, and security. These decisions will help you understand what direction to take. Finally, we’ll share some lessons learned with the pieces of our architecture worked well and rant about those which didn’t. No deep Hadoop knowledge is necessary, architect or executive level.
Creating a Data Driven Organization - StampedeCon 2016StampedeCon
Companies today are all focused on finding new consumption models to better utilize the data they produce. This presentation will provide insights and best practices for creating the organization and sponsorship necessary to set the foundation for success.
For this session, Dan will provide an overview of the process and methodologies he employs to establish and sustain a Data Driven Culture. Key topics will include:
Data Driven Culture
Executive Sponsorship
Organizational Structure – Collaboration Hubs and Bi-Modal Analytics
Role of Hadoop and Big Data as Part of Data Driven Culture
Using The Internet of Things for Population Health Management - StampedeCon 2016StampedeCon
The Internet of (Human) Things is just beginning to take shape. The human body is an inexhaustible source of data about personal health, and the healthcare industry is just beginning to scratch the surface of the potential insights and value that will come from that data. While much of healthcare traditionally focuses on the episodic delivery of services, the Affordable Care Act is pushing healthcare providers, payers, and self-funded employer groups to look at ways to proactively encourage healthy behaviors. Providing personal health devices as a way to promote individual health is one way that healthcare is beginning to take advantage of IoT technologies. This session provides insight into how IoT is being leveraged in population health management through a solution jointly delivered by Amitech Solutions and Big Cloud Analytics. Attendees will learn how Hadoop is being used to gather personal device from various vendors, integrate and analyze that information, differentiate trends across regional and cultural diversity, and provide personal recommendations and insights into health risks. This session presents one important way the healthcare industry is leveraging IoT.
Turn Data Into Actionable Insights - StampedeCon 2016StampedeCon
At Monsanto, emerging technologies such as IoT, advanced imaging and geo-spatial platforms; molecular breeding, ancestry and genomics data sets have made us rethink how we approach developing, deploying, scaling and distributing our software to accelerate predictive and prescriptive decisions. We created a Cloud based Data Science platform for the enterprise to address this need. Our primary goals were to perform analytics@scale and integrate analytics with our core product platforms.
As part of this talk, we will be sharing our journey of transformation showing how we enabled: a collaborative discovery analytics environment for data science teams to perform model development, provisioning data through APIs, streams and deploying models to production through our auto-scaling big-data compute in the cloud to perform streaming, cognitive, predictive, prescriptive, historical and batch analytics@scale, integrating analytics with our core product platforms to turn data into actionable insights.
Enhanced Enterprise Intelligence with your personal AI Data Copilot.pdfGetInData
Recently we have observed the rise of open-source Large Language Models (LLMs) that are community-driven or developed by the AI market leaders, such as Meta (Llama3), Databricks (DBRX) and Snowflake (Arctic). On the other hand, there is a growth in interest in specialized, carefully fine-tuned yet relatively small models that can efficiently assist programmers in day-to-day tasks. Finally, Retrieval-Augmented Generation (RAG) architectures have gained a lot of traction as the preferred approach for LLMs context and prompt augmentation for building conversational SQL data copilots, code copilots and chatbots.
In this presentation, we will show how we built upon these three concepts a robust Data Copilot that can help to democratize access to company data assets and boost performance of everyone working with data platforms.
Why do we need yet another (open-source ) Copilot?
How can we build one?
Architecture and evaluation
Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...John Andrews
SlideShare Description for "Chatty Kathy - UNC Bootcamp Final Project Presentation"
Title: Chatty Kathy: Enhancing Physical Activity Among Older Adults
Description:
Discover how Chatty Kathy, an innovative project developed at the UNC Bootcamp, aims to tackle the challenge of low physical activity among older adults. Our AI-driven solution uses peer interaction to boost and sustain exercise levels, significantly improving health outcomes. This presentation covers our problem statement, the rationale behind Chatty Kathy, synthetic data and persona creation, model performance metrics, a visual demonstration of the project, and potential future developments. Join us for an insightful Q&A session to explore the potential of this groundbreaking project.
Project Team: Jay Requarth, Jana Avery, John Andrews, Dr. Dick Davis II, Nee Buntoum, Nam Yeongjin & Mat Nicholas
Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...Subhajit Sahu
Abstract — Levelwise PageRank is an alternative method of PageRank computation which decomposes the input graph into a directed acyclic block-graph of strongly connected components, and processes them in topological order, one level at a time. This enables calculation for ranks in a distributed fashion without per-iteration communication, unlike the standard method where all vertices are processed in each iteration. It however comes with a precondition of the absence of dead ends in the input graph. Here, the native non-distributed performance of Levelwise PageRank was compared against Monolithic PageRank on a CPU as well as a GPU. To ensure a fair comparison, Monolithic PageRank was also performed on a graph where vertices were split by components. Results indicate that Levelwise PageRank is about as fast as Monolithic PageRank on the CPU, but quite a bit slower on the GPU. Slowdown on the GPU is likely caused by a large submission of small workloads, and expected to be non-issue when the computation is performed on massive graphs.
The Building Blocks of QuestDB, a Time Series Databasejavier ramirez
Talk Delivered at Valencia Codes Meetup 2024-06.
Traditionally, databases have treated timestamps just as another data type. However, when performing real-time analytics, timestamps should be first class citizens and we need rich time semantics to get the most out of our data. We also need to deal with ever growing datasets while keeping performant, which is as fun as it sounds.
It is no wonder time-series databases are now more popular than ever before. Join me in this session to learn about the internal architecture and building blocks of QuestDB, an open source time-series database designed for speed. We will also review a history of some of the changes we have gone over the past two years to deal with late and unordered data, non-blocking writes, read-replicas, or faster batch ingestion.
06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Data and AI
Round table discussion of vector databases, unstructured data, ai, big data, real-time, robots and Milvus.
A lively discussion with NJ Gen AI Meetup Lead, Prasad and Procure.FYI's Co-Found
06-04-2024 - NYC Tech Week - Discussion on Vector Databases, Unstructured Data and AI
Discussion on Vector Databases, Unstructured Data and AI
https://www.meetup.com/unstructured-data-meetup-new-york/
This meetup is for people working in unstructured data. Speakers will come present about related topics such as vector databases, LLMs, and managing data at scale. The intended audience of this group includes roles like machine learning engineers, data scientists, data engineers, software engineers, and PMs.This meetup was formerly Milvus Meetup, and is sponsored by Zilliz maintainers of Milvus.
Adjusting OpenMP PageRank : SHORT REPORT / NOTESSubhajit Sahu
For massive graphs that fit in RAM, but not in GPU memory, it is possible to take
advantage of a shared memory system with multiple CPUs, each with multiple cores, to
accelerate pagerank computation. If the NUMA architecture of the system is properly taken
into account with good vertex partitioning, the speedup can be significant. To take steps in
this direction, experiments are conducted to implement pagerank in OpenMP using two
different approaches, uniform and hybrid. The uniform approach runs all primitives required
for pagerank in OpenMP mode (with multiple threads). On the other hand, the hybrid
approach runs certain primitives in sequential mode (i.e., sumAt, multiply).
Techniques to optimize the pagerank algorithm usually fall in two categories. One is to try reducing the work per iteration, and the other is to try reducing the number of iterations. These goals are often at odds with one another. Skipping computation on vertices which have already converged has the potential to save iteration time. Skipping in-identical vertices, with the same in-links, helps reduce duplicate computations and thus could help reduce iteration time. Road networks often have chains which can be short-circuited before pagerank computation to improve performance. Final ranks of chain nodes can be easily calculated. This could reduce both the iteration time, and the number of iterations. If a graph has no dangling nodes, pagerank of each strongly connected component can be computed in topological order. This could help reduce the iteration time, no. of iterations, and also enable multi-iteration concurrency in pagerank computation. The combination of all of the above methods is the STICD algorithm. [sticd] For dynamic graphs, unchanged components whose ranks are unaffected can be skipped altogether.
Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...
The Search for a New Visual Search Beyond Language - StampedeCon AI Summit 2017
1. This is your cover.
Insert an amazing image and align it
with this grey rectangle. Please use the
red, patterned, Shutterstock background
for internal presentations only.
The search for a new visual search
Mike Ranzinger, Senior Research Engineer @ Shutterstock
3. • We’re going to explore a new technology that was just released to beta called “Composition
Aware Search”
• This involves some key technologies:
•Convolutional Neural Nets (Vision and NLP)
•Discriminative Localization
•Multi-modal Embeddings
•Dimensionality Reduction
•Inverted Multi-Index
• Yes, between this presentation, and our publicly shared white paper, you should be able to
implement this yourself
•(non-commercially of course).
Outline
6. An image can help provide visual
interest to your written content.
Insert an image and align
it with this grey rectangle.
Domain mismatch
• The saying is: “A picture is worth a thousand
words”
• Our average query length is 2 words
• Sometimes it’s hard to describe exactly what
you’re looking for
• Our users are accustomed to looking
through multiple pages of results to find
what they
were looking for
8. • Common problem: I have picture X without a license, and I need to get a
license for it
•Perhaps you saw it on social media, and you wanted to share it more officially
• My toy problem: I took this bad picture, find me a good one!
• We don’t use words, at all. We communicate through pixels.
Reverse image search
9. My bike
How does it work?
Trained CNN
Fixed length
vectorTrained CNN
Our bike images
Maximum inner product
search between our
collection, and the
query vector
OurCollection
10. • We have a vision model that can produce
an N-dimensional vector for a given image.
• Train a language model that maps a query
to the vector of the downloaded image.
• Training set: Query to download pairs.
Multimodal embedding / query language models
Lemur on rock
11. • Kiros et. al. “Unifying visual-semantic embeddings with multimodal
neural language models”
• Trained using “Triplet Loss”
•Let 𝑓(𝑥) be the L2 normalized output of the vision model on image 𝑥
•Let 𝑔(𝑞) be the L2 normalized output of the language model on query 𝑞
•Let 𝑞' be the query corresponding to image 𝑥'
•Let 𝑚 be some margin 0 < 𝑚 < 2
• 𝐿 = max 0, 𝑓 𝑥2 ∘ 𝑔 𝑞' − 𝑓 𝑥' ∘ 𝑔 𝑞' + 𝑚
• In words, the dot product between a query and it’s corresponding
image (green) should be greater than the query and some unrelated
image (red) by some margin 𝑚.
Multimodal embedding
12. • We train the vision model first
• Next, we train the language model.
•We don’t backprop gradients though the vision model because it degrades
it
• Once we’ve finished training the language model, we can search for
images given a query using MIPS, the same way that we did with reverse
image search.
Multimodal embedding
14. • Here’s an example of a “fully convolutional” neural network.
• A fully convolutional network is typically a series of convolutions and
downsampling operations that ends with a global average pooling operation.
• The GAP reduces the final feature maps down to a single vector (one value per
feature map).
• We call a position (y, x) in the final feature maps a “spatial feature vector”
Spatial feature vectors
15. • Like the feature vector produced by the global average pool,
spatial vectors also encode information in the same
embedding space.
• Importantly, these vectors tend to encode more localized
information based on the receptive field of the given neuron.
• We exploit these vectors to build out CAS
Spatial feature vectors
16. • Zhou et. al. introduced a very important paper
titled “Learning Deep Features for Discriminative
Localization”
• They introduce the concept of “Class Activation
Maps” (CAM), which is effectively a heatmap of
the classification strength for each output
position before the GAP, for a given class.
Discriminative localization
17. • Let 𝑓6 𝑦, 𝑥 be the activation of unit 𝑘 at position (y, x) of the last
convolutional layer
• 𝐹6 =
'
:,;
∑ 𝑓6 𝑦, 𝑥:,;
•The result of the GAP for unit 𝑘
• For a given class 𝑐, the input to the softmax, 𝑆@ = ∑ 𝑤6
@
6 𝐹6
•In words, the dot product between the GAP features, and the learned
vector for the given class, where 𝑤6
@
is the weight for class 𝑐 for unit 𝑘
• Let 𝑀@ 𝑦, 𝑥 = ∑ 𝑤6
@
6 𝑓6 𝑦, 𝑥 be the class activation map, or in
words, the importance of spatial position (y,x) for the classification
of class c.
•I’d recommend reading the paper to see the full derivation
Discriminative localization
18. CAM for highest probability guess, which is “meerkat” with probability 40%.
What it looks like for us
20. • Recall that the output of the GAP is
• 𝐹6 =
'
:,;
∑ 𝑓6 𝑦, 𝑥:,;
• What if, instead of needing class 𝑐, we instead use 𝐹6 as the target
• 𝑀@ 𝑦, 𝑥 = ∑ 𝐹66 𝑓6 𝑦, 𝑥
• Basically, this tells us how close a given spatial vector is to the average
vector. One way to interpret this is, “how salient is the spatial vector to the
classification”.
Auto-saliency
21. Note that “lemur” isn’t actually a class that the network was trained against. The
closest class neighbors are meerkat and koala.
Auto-saliency
25. • Why is this important?
•It allows us to visualize how the network behaves on inputs for classes that it
wasn’t explicitly trained on.
• The idea that this works also reveals an open problem for us:
•In order for the salient vectors to emerge, the non-salient regions of the image
must either try to align themselves in the same direction as the salient vector
•Dilation
•Or, the non-salient regions must reduce their magnitude so not to bias the
salient vector
Auto-saliency
26. • We have now seen how we can use CAMs, as well as the GAP vectors
themselves to guide the heatmaps.
• Finally, we can look back at the language model we trained earlier.
• 𝐿 = max 0, 𝑓 𝑥2 ∘ 𝑔 𝑞' − 𝒇 𝒙 𝟏 ∘ 𝒈 𝒒 𝟏 + 𝑚
•The language model learns to match the direction of the GAP
•In effect, we can use the language model to generate the class weights for the CAM
technique on the fly.
•I think it’s neat to interpret the language model as a low-rank approximator of
the (potentially infinite) classification weight matrix.
Language models as discriminators
27. Composition aware search - overview
Spatial IndexCollect
This texture is also an anchor,
with position, size, and query
image.
Lamp
Lamp and Chair are called
“anchors”, which have both
a position and a query string.
Chair
VisionModel
Language Model
28. • Vision Model
•We are using a variant of the Inception v3 paper by Szegedy et. al. titled
“Rethinking the Inception Architecture for Computer Vision”
•Notable differences:
•We are not using batch normalization
•We are using ELU non-linearities instead of ReLUs.
• Language Model
•We tried to be fancy and use cool tech such as character-models and LSTMs
•The character LSTMs massively overfit on us
•So, we used words, and dropped recurrency altogether in favor of a simpler
convolutional language model as described by Collobert et. al. in “Natural
Language Processing (Almost) from Scratch”
Models
29. • Let’s look at the query formulation for this, starting
with the simple case.
• Let 𝑆(𝑖) be the score for image 𝑖
• Let 𝐐 be the set of anchors, and 𝐪L be the 𝑗-th
anchor L2 normalized (column) vector
• Let 𝐕O be the set of spatial vectors in image 𝑖, and 𝐯OQ
be the 𝑝-th spatial L2 normalized (column) vector
• Let 𝑤LQ be a positional weight applied to position 𝑝
based on the position of query 𝑗
The search problem
𝑆(𝑖) =
1
𝐐
T max
UVQVW
𝑤LQ 𝐯OQ
⏉
𝐪L
𝐐
L
30. Take the average over the anchors.
The search problem
𝑆 𝑖 =
1
𝐐
T max
UVQVW
𝑤LQ 𝐯OQ
⏉
𝐪L
𝐐
L
We only care about the largest
weighted similarity score. This
gives us a single score per
query anchor for an image.
We use this to weight the similarity
score based on the relative position
of the anchor to the spatial vector.
Take the average over the anchors.
32. The search problem
• Using the above definition, the size of the index is
defined by the following variables:
• 𝐶 - The size of the collection
• 𝐷 - The dimensionality of the spatial vectors
• 𝑃 - The number of spatial positions
• For our (beta) production offering, we have:
• 𝐶 = 10,000,000
• 𝐷 = 256
• 𝑃 = 64, we use 8 rows and 8 columns
• Requires about 611 GB of space to store the index.
• Algorithm complexity is also 𝑂(𝐶 ⋅ 𝐐 ⋅ 𝑃 ⋅ 𝐷),
which is, a lot.
𝑆(𝑖) =
1
𝐐
T max
UVQVW
𝑤LQ 𝐯OQ
⏉
𝐪L
𝐐
L
33. Visualizing concepts
Using PCA, we can visualize how the concepts are arranged based
on the 2 principal directions.
Cat
Background
36. Visualizing concepts – reduction methods
• t-SNE was superior at disentangling concepts on a 2d plane
•Manifold learning technique
•Popular technique for data visualization
• PCA was still able to do a decent job
•Linear
• We use PCA because embedding a new point is efficiently
computed with a single GEMM
37. The search problem (part 2)
• Since we are performing a dimensionality
reduction on the spatial vectors for each image,
let’s re-define the search problem.
• Let 𝐁O ∈ ℝd×f be the orthonormal basis of the
PCA for image 𝑖 such that we preserve 𝑁
dimensions, and 𝑁 ≤ 𝐷.
• Let 𝐝OQ = 𝐁O 𝐯OQbe the reduced dimensionality
spatial vector for image 𝑖 at position 𝑝.
𝑆 𝑖 =
1
𝐐
T max
UVQVW
𝑤LQ 𝐝OQ
⏉
𝐁O 𝐪L
𝐐
L
38. The search problem (part 2)
Then compute the dot product
between the two vectors in the
subspace.
Project the query vector into the
reduced dimensionality subspace
for the image.
𝑆 𝑖 =
1
𝐐
T max
UVQVW
𝑤LQ 𝐝OQ
⏉
𝐁O 𝐪L
𝐐
L
39. The search problem (part 2)
Naïve definition
• Requires 𝐶𝐷 𝑃 storage space
•611 GB for 10mil images
• Computation 𝑂(𝐶 ⋅ 𝐐 ⋅ 𝑃 ⋅ 𝐷)
•In practice,
𝑃 = 64, 𝐷 = 256, so 𝑃 ⋅ 𝐷 = 16384
𝑆(𝑖) =
1
𝐐
T max
UVQVW
𝑤LQ 𝐯OQ
⏉
𝐪L
𝐐
L
𝑆 𝑖 =
1
𝐐
T max
UVQVW
𝑤LQ 𝐝OQ
⏉
𝐁O 𝐪L
𝐐
L
New definition
• Requires 𝐶𝑁 𝐷 + 𝑃 storage space
• 𝑁 ≈ 4
•48 GB for 10mil images
•12.7x reduction
• 𝑂 𝐶 ⋅ 𝐐 𝑁 𝐷 + 𝑃
• 𝑁 𝐷 + 𝑃 = 1280
•12.8x reduction
40. The search problem (part 3)
• Now the current computational complexity is:
• 𝑂 𝐶 ⋅ 𝐐 𝑁 𝐷 + 𝑃
• Importantly, this is still intractable because it still processes
every image in the collection.
•Users typically don’t want to wait roughly 7 seconds for a
response
• Our best bet is to formulate the problem such that we only
process a tiny fraction of 𝐶
41. Inverted index
• Construction:
•Select codebook size, 𝐖
•Find the 𝐖 centroids of 𝐶 using a K-means like process
•Each of these centroids are called “codewords”
•Assign each vector in 𝐂 to it’s nearest vector in 𝐖
• Inference:
•Find the 𝑘 nearest codewords in 𝐖 to 𝐪
•Either return all of the vectors in the 𝑘 codewords, or perhaps
find the 𝑘′ nearest vectors to 𝐪 within the codewords.
42. Inverted multi-index
• Introduced by Babenko and Lempitsky at CVPR 2012
• This technique combines Product Quantization with
Inverted Indices
• Construction:
•Split your collection 𝐶 into 𝑁 partitions, typically 𝑁 = 2
•For each partition 𝑀, find 𝐖 cluster centers, as with
the inverted index
•For each vector in 𝐶, assign it to the nearest codeword
in each partition independently.
•This forms a Cartesian product of the codebooks,
such that the full codebook is essentially size 𝐖 d
• Paper: Inverted Multi-Index
Dims 1 → 𝑟 Dims 𝑟 + 1 → 𝐷
43. Inverted multi-index
• Inference:
•Sort the codebooks in each partition 𝑀 based on
distance to 𝐪′
•Traverse the 𝑁 codebooks by visiting the nearest
codeword 𝑚 defined by the sum of the distances for
each 𝐪′ to each 𝐦.
• I strongly recommend reading the paper for this one.
It’s hard to explain on a slide.
44. Inverted multi-index
Source: “The Inverted Multi-Index” by Artem Babenko and Victor Lempitsky
Visualization of a set of datapoints, and their respective clusters.
45. The inverted multi-index for
CAS
• For the most part, we use the basic formulation of the IMI
•We use 𝐖 = 10000, which results in 100-million possible codewords
• Except:
•Each image has 𝑃 spatial vectors associated with it, so we assign each
spatial vector to a cluster independently
•This is actually the main reason we use the IMI over the regular
inverted index, because we effectively have 𝑃 ⋅ 𝐶 vectors to index,
and the inverted index doesn’t scale as well.
•The paper primarily addresses 𝐿2 distance, but we use cosine distance
•Scale the codebook vectors by
d
d
such that the magnitude of any
set of codewords 𝑚', 𝑚2, ⋯ , 𝑚t = 1
46. The inverted multi-index for
CAS
• We expand clusters for each query term until we reach a fixed
number of images
• We then take the set union of expansions for each query term,
and run the previously defined scoring function.
• We look for about 5k images per anchor, so we typically only
rank between 0.05% and 0.15% of the collection.
47. Spatial-semantic image search
by visual feature synthesis
Mai et. al. introduced the above titled paper at CVPR 2017
Credit: Mai et. Al.
Spatial-semantic image search by visual feature synthesis
48. Mai et. al. paper
• Joint effort between Portland State University and Adobe
Research
• Their problem space is very similar to CAS
• Key Differences:
•Their language model learns to map all of the non-uniformly
sized anchors to a single feature vector, and then search
proceeds like a standard nearest neighbors query.
•They train their models using a dataset with object
localization information (COCO)
• Basically, if your data has localized labels, their approach is
very compelling.
49. Levels Of Supervision
Unsupervised
• No labeled data
• GANs, VAEs, etc.
Where I want to be.
LeCun thinks so too.
Semi-supervised
• Some labeled data
What best
leverages
Shutterstock’s
data
Supervised
• Classification labels
• ILSVRC, etc.
Where CAS is
currently
Very supervised
• Classification and
localization labels,
sometimes even
pixel level
segmentation.
• COCO, etc.
Mai et. al.
50. Challenges to the current system
• The global average pool encourages a couple different bad behaviors:
•It “dilates” the salient regions of the image, such that neurons that
are near the salient concept adopt the salient vectors instead of
representing the primary concept of their own receptive field
•It creates a hierarchy of vector magnitudes, such that salient concepts
have much larger magnitude than less salient concepts. This can allow
the network to learn less-robust representations of the non-salient
image patches.
51. This is your closing slide..
Insert an amazing image and align it with this
grey rectangle for a dramatic transition.
Feel free to change the copy to white should
want it to show up better against the image.
Thank you!
Mike Ranzinger
mranzinger@shutterstock.com
www.Shutterstock.com/labs/compositionsearch