ODSC West 2019

Training | Data Visualization | Open Source Data Science | Intermediate

Have you ever wondered about how those data scientists at Facebook and LinkedIn make friend recommendations? Or how epidemiologists track down patient zero in an outbreak? If so, then this tutorial is for you. In this tutorial, we will use a variety of datasets to help you understand the fundamentals of network thinking, with a particular focus on constructing, summarizing, and visualizing complex networks.

This tutorial is for Pythonistas who want to understand relationship problems – as in, data problems that involve relationships between entities. Participants should already have a grasp of for loops and basic Python data structures (lists, tuples and dictionaries). By the end of the tutorial, participants will have learned how to use the NetworkX package in the Jupyter environment, and will become comfortable in visualizing large networks using Circos plots. Other plots will be introduced as well…more details

Training | Deep Learning | Open Source Data Science | Beginner-Intermediate

Relatively obscure a few short years ago, Deep Learning is ubiquitous today across data-driven applications as diverse as machine vision, natural language processing, and super-human game-playing.

This Deep Learning primer brings the revolutionary machine-learning approach behind contemporary artificial intelligence to life with interactive demos featuring TensorFlow 2.0, the major, cutting-edge revision of the world’s most popular Deep Learning library.

To facilitate an intuitive understanding of Deep Learning’s artificial-neural-network foundations, the essential theory will be introduced visually and pragmatically. Paired with tips for overcoming common pitfalls and hands-on Python code run-throughs provided in straightforward Jupyter notebooks, this foundational knowledge empowers you to build powerful state-of-the-art Deep Learning models…more details

Training | Machine Learning | Open Source Data Science | Beginner-Intermediate

We’ll start with the basics of machine learning on Apache Spark: when to use it, how it works, and how it compares to all of your other favorite data science tooling.

You’ll learn to use Spark (with Python) for statistics, modeling, scoring (inference), and model tuning. But you’ll also get a peek behind the APIs: see why the pieces are arranged as they are, how to get the most out of the docs, open source ecosystem, third-party libraries, and solutions to common challenges…more details

Training | Machine Learning | Intermediate

Many industry segments have been grappling with fast data (high-volume, high-velocity data). The enterprises in these industry segments need to process this fast data just in time to derive insights and act upon it quickly. Such tasks include but are not limited to enriching data with additional information, filtering and reducing noisy data, enhancing machine learning models, providing continuous insights on business operations, and sharing these insights just in time with customers. In order to realize these results, an enterprise needs to build an end-to-end data processing system, from data acquisition, data ingestion, data processing, and model building to serving and sharing the results. This presents a significant challenge, due to the presence of multiple messaging frameworks and several streaming computing frameworks and storage frameworks for real-time data. In the proposed tutorial, we shall lead a journey through the landscape of state-of-the-art systems for each stage of an end-to-end data processing pipeline, messaging frameworks, streaming computing frameworks, storage frameworks for real-time data, and more…more details

Training | Deep Learning | Intermediate

Deep learning is the newest area of machine learning and has become ubiquitous in predictive modeling. The complex, brainlike structure of deep learning models is used to find intricate patterns in large volumes of data. These models have heavily improved the performance of general supervised models, time series, speech recognition, object detection and classification, and sentiment analysis.

Factorization machines are a relatively new and powerful tool for modeling high-dimensional and sparse data. Most commonly they are used as recommender systems by modeling the relationship between users and items. For example, factorization machines can be used to recommend your next Netflix binge based on how you and other streamers rate content.

In this session, participants will use recurrent neural networks to analyze sequential data and improve the forecast performance of time series data, and use convolutional neural networks for image classification. Participants will also use an intelligent search method to efficiently tune the hyperparameters of both deep learning models…more details

Training | Deep Learning | Machine Learning | Beginner – Intermediate

In this workshop you will discover how machines can learn complex behaviors and anticipatory actions. Using this approach autonomous helicopters fly aerobatic maneuvers and even the GO world champion was beaten with it. A training dataset containing the “right” answers is not needed, nor is “hard-coded” knowledge. The approach is called “reinforcement learning” and is almost magical.

Using TF-Agents on top of TensorFlow 2.0 we will see how a real-life problem can be turned into a reinforcement learning task. In an accompanying Python notebook, we implement – step by step – all solution elements, highlight the design of Google’s newest reinforcement learning library, point out the role of neural networks and look at optimization opportunities…more details

Training | Machine Learning | Open Source Data Science | All Levels

Scikit-learn is a machine learning library in Python, that has become a valuable tool for many data science practitioners. This training will cover some advanced topics in using scikit-learn, such as how to perform out-of-core learning with scikit-learn and how to speed up parameter search. We’ll also cover how to build your own models or feature extraction methods that are compatible with scikit-learn, which is important for feature extraction in many domains. We will see how we can customize scikit-learn even further, using custom methods for cross-validation or model evaluation.

This workshop assumes familiarity with Jupyter notebooks and basics of pandas, matplotlib and numpy. It also assumes experience using scikit-learn and familiarity with the API

Training | Kickstarter | Machine Learning | All Levels

Scikit-learn is a machine learning library in Python, that has become a valuable tool for many data science practitioners. This training will cover some of the more advanced aspects of scikit-learn, such as building complex machine learning pipelines, advanced model evaluation, feature engineering and working with imbalanced datasets. We will also work with text data using the bag-of-word method for classification.

This workshop assumes familiarity with Jupyter notebooks and basics of pandas, matplotlib and numpy. It also assumes some familiarity with the API of scikit-learn and how to do cross-validations and grid-search with scikit-learn.

Training | Machine Learning | Open Source Data Science | Beginner – Intermediate

Scikit-learn is a machine learning library in Python, that has become a valuable tool for many data science practitioners. This talk will cover some of the more advanced aspects of scikit-learn, such as building complex machine learning pipelines, model evaluation, parameter search, and out-of-core learning. Apart from metrics for model evaluation, we will cover how to evaluate model complexity, and how to tune parameters with grid search, randomized parameter search, and what their trade-offs are. We will also cover out of core text feature processing via feature hashing.

Training | Kickstarter | Machine Learning | All Levels

Machine learning has become an indispensable tool across many areas of research and commercial applications. From text-to-speech for your phone to detecting the Higgs boson, machine learning excells at extracting knowledge from large amounts of data. This talk will give a general introduction to machine learning, as well as introduce practical tools for you to apply machine learning in your research. We will focus on one particularly important subfield of machine learning, supervised learning. The goal of supervised learning is to “”learn”” a function that maps inputs x to an output y, by using a collection of training data consisting of input-output pairs. We will walk through formalizing a problem as a supervised machine learning problem, creating the necessary training data and applying and evaluating a machine learning algorithm. The talk should give you all the necessary background to start using machine learning yourself.

Training | Data Visualization | Intermediate

How can an idea from an 18th century Presbyterian minister be used to estimate the mass density function of galaxies across the Universe? How can a marketing segmentation algorithm protect astronauts traveling to Mars from certain death? How does a Formula 1 race from then 1950’s inspire one of the greatest data science use cases for the Internet of Things? How can a violation of the triangle inequality theorem in mathematics lead to a cure for cancer? This workshop will answer these questions, and more, by presenting several examples of one of the key aptitudes of successful data science practice, which is adaptability. In particular, I will present several well known algorithms (including some that we would not even call “algorithms”) that may have been adopted for specific use cases or applied in specific business domains…more details

Training | Kickstarter | Data Visualization | Beginner – Intermediate

Markdown Primer (45 minutes): Structure Documents with Sections and Subsections, Formatting Text, Creating Ordered and Unordered Lists, Making Links, Number Sections, Include Table of Contents
Integrate R Code (30 minutes): Insert Code Chunks, Hide Code, Set Chunk Options, Draw Plots, Speed Up Code with Caching
Build RMarkdown Slideshows (20 minutes): Understand Slide Structure, Create Sections, Set Background Images, Include Speaker Notes, Open Slides in Speaker Mode
Develop Flexdashboards (30 minutes): Start with the Flexdashboard Layout, Design Columns and Rows, Use Multiple Pages, Create Social Sharing, Include Code
Shiny InputsDrop DownsTextRadioChecks
Shiny OutputsTextTablesPlots
Reactive Expressions
HTML WidgetsInteractive PlotsInteractive MapsInteractive Tables
Shiny Layouts UI and Server Files User Interface

Training | Data Visualization | Open Source Data Science | Intermediate

Markdown Primer (45 minutes): Structure Documents with Sections and Subsections, Formatting Text, Creating Ordered and Unordered Lists, Making Links, Number Sections, Include Table of Contents
Integrate R Code (30 minutes): Insert Code Chunks, Hide Code, Set Chunk Options, Draw Plots, Speed Up Code with Caching
Build RMarkdown Slideshows (20 minutes): Understand Slide Structure, Create Sections, Set Background Images, Include Speaker Notes, Open Slides in Speaker Mode
Develop Flexdashboards (30 minutes): Start with the Flexdashboard Layout, Design Columns and Rows, Use Multiple Pages, Create Social Sharing, Include Code
Shiny InputsDrop DownsTextRadioChecks
Shiny OutputsTextTablesPlots
Reactive Expressions
HTML WidgetsInteractive PlotsInteractive MapsInteractive Tables
Shiny Layouts UI and Server Files User Interface

Training | Machine Learning | Open Source Data Science | All Levels

Modern statistics has become almost synonymous with machine learning, a collection of techniques that utilize today’s incredible computing power. This two-part course focuses on the available methods for implementing machine learning algorithms in R, and will examine some of the underlying theory behind the curtain. We start with the foundation of it all, the linear model and its generalization, the glm. We look how to assess model quality with traditional measures and cross-validation and visualize models with coefficient plots. Next we turn to penalized regression with the Elastic Net. After that we turn to Boosted Decision Trees utilizing xgboost. Attendees should have a good understanding of linear models and classification and should have R and RStudio installed, along with the `glmnet`, `xgboost`, `boot`, `ggplot2`, `UsingR` and `coefplot` packages…more details

Training | Kickstarter | Machine Learning | All Levels

Modern statistics has become almost synonymous with machine learning, a collection of techniques that utilize today’s incredible computing power. This two-part course focuses on the available methods for implementing machine learning algorithms in R, and will examine some of the underlying theory behind the curtain. We start with the foundation of it all, the linear model and its generalization, the glm. We look how to assess model quality with traditional measures and cross-validation and visualize models with coefficient plots. Next we turn to penalized regression with the Elastic Net. After that we turn to Boosted Decision Trees utilizing xgboost. Attendees should have a good understanding of linear models and classification and should have R and RStudio installed, along with the `glmnet`, `xgboost`, `boot`, `ggplot2`, `UsingR` and `coefplot` packages…more details

Training | Deep Learning | Open Source Data Science | Beginner – Intermediate

In this session, we will cover how to create Deep Neural Networks using the PyTorch framework on a variety of examples. The material will range from beginner – understanding what is going on “”under the hood””, coding the layers of our networks, and implementing backpropagation – to more advanced material on RNNs,CNNs, LSTMs, & GANs.

Attendees will leave with a better understanding of the PyTorch framework. In particular, how it differs from Keras and Tensorflow. Furthermore, a link to a clean documented GitHub repo with the solutions of the examples covered will be provided.

Training | Open Source Data Science | DevOps & Managment | Beginner – Intermediate

The intention of this VERY hands on workshop is to get you introduced and playing with some of the great features you never knew about in PostgreSQL. You know, and probably already love, PostgreSQL as your relational database. We will show you how you can forget about using ElasticSearch, MongoDB, and Redis for a broad array of use cases. We will add in some nice statistical work with R embedded in PostgreSQL. Finally we will bring this all together using the gold standard in spatial databases, PostGIS. Unless you have a specialized use case, PostgreSQL is the answer. The session will be very hands on with plenty of interactive exercises.

By the end of the workshop participants will leave with hands on experience doing Spatial Analysis, JSON search, Full Text Search , Using R for stored procedures and functions, All in PostgreSQL…more details

Training | Deep Learning | Intermediate

Reinforcement learning considers the problem of learning to act and is poised to power next generation AI systems, which will need to go beyond input-output pattern recognition (even if such simpler AI has sufficed for speech, vision, machine translation) and will have to generate intelligent behavior. Example application domains include robotics, marketing, dialogue, HVAC, optimizing healthcare and supply chains.

In this tutorial we will cover the foundations of Deep RL (including, but not limited to: CEM, DQN, TRPO, PPO, SAC) as well as dive into the specifics of some of the main success stories and provide perspective on where the field is headed.

To get the most out of this tutorial, the audience is assumed to have basic familiarity with neural networks, optimization, probability…more details

Training | Machine Learning | Data Visualization | Beginner-Intermediate

Literate computing weaves a narrative directly into an interactive computation. Text, code, and results are combined into a narrative that relies equally on textual explanations and computational components. Insights are extracted from data using computational tools. These insights are communicated to an audience in the form of a narrative that resonates with the audience. Literate computing lends itself to the practice of reproducible research. One may re-run the analyses; run the analyses with new data sets; modify the code for other purposes.

This workshop will take one through the steps associated with literate computing: data retrieval; data curation; model construction, evaluation, and selection; and reporting. Particular attention will be paid to reporting, i.e., building a narrative. Examples will be presented demonstrating how one might generate multiple output formats (e.g., HTML pages, presentation slides, PDF documents) starting with the same code base…more details

Training | Deep Learning | AI for Engineers | Intermediate-Advanced

We will build and tweak several vision classifiers together starting with perceptrons and building up to transfer learning and convolutional neural networks. We will investigate practical implications of tweaking loss functions, gradient descent algorithms, network architectures, data normalization, data augmentation and so on. This class is super hands on and practical and requires no math or experience with deep learning.

Training | Deep Learning | Intermediate-Advanced

This is a very hands on introduction to LSTMs in Keras and TensorFlow. We will build a language classifier, generator and a translating sequence to sequence model. We will talk about debugging models and explore various related architectures like GRUs, Bidirectional LSTMs, etc. to see how well they work.

Training | DevOps & Managment | Intermediate

Humans evolved to tell stories. In fact, we evolved to rely on story for our most important learning. Some argue story was more important for the survival of the species than opposable thumbs.

In this half-day workshop, learn how to take a step back from your data and think like a storyteller. Learn some key ideas and techniques to turn your numbers into a narrative – to make a compelling story that will have an impact on your audience. We will cover practical, actionable ideas that will make your next effort at communicating with data much more powerful…more details

Talks | Machine Learning | DevOps & Managment | Beginner-Intermediate

We are in the age of data. In recent years, many companies have already started collecting large amounts of data about their business. Many other companies are starting now.

However, you know that before you can train any decent supervised model you need ground truth data. Usually, supervised ML models are trained on old data records that are already somehow labeled. And this is the ugly truth: before proceeding with any model training, any classification problem definition, or any further enthusiasm in gathering data, you need a sufficiently large set of correctly labeled data records to describe your problem. And data labeling – especially in a sufficiently large amount – is … expensive…more details

Talk | Machine Learning | DevOps & Managment | Intermediate

Data resources in the Earth Sciences range from kilobytes to terabytes, with a range of user communities, technical competence, and scientific uses. As interdisciplinary research continues to grow there is a significant need to increase usability and discoverability, to assign credit to data generators, developers and to provide people with the tools and knowledge they need to help address some of society’s most pressing issues. This presentation showcases an approach that uses neo4j’s graph database, open source APIs, the use of persistent identifiers such as ORCIDs and DOIs, and crowd sourcing to help connect disciplines within the Earth Sciences to undertake science for a new century.

Talk | Open Source Data Science | Machine Learning | Beginner

Cloud storage brings great flexibility in management and cost-efficiency to data scientists, but also introduces new challenges related to data accessibility and data locality for machine learning applications. For instance, when the input data is stored in a remote cloud storage like AWS S3 or Azure blob storage, direct data access is often slow and expensive; but manually moving data to the training clusters can be time-consuming, complicated and often require data engineering or ETL pipelines.

This session is designed for data scientists or data engineers who work with remote and possibly multiple data sources in hybrid or multi-cloud environments…more details

Talk | Intermediate

Given the verbosity with which modern software produces logs – it is useful to have many dimensions for filtering when looking for specific content. Often, groups of log messages will relate to a general software concept (e.g. security, resource utilization, database access, etc.) and it can be useful to examine these messages as a group or to pinpoint messages that fall within the intersection of one or more of these concepts. Here we describe our approach to classifying previously unseen log messages into these software concept categories. To handle the large domain-specific vocabulary used by log messages we augmented the “continuous bag of words” (CBOW) embedding training process with an additional semi-supervised training step in which we create a “concept vector”. This vector of concept terms was produced by interrogating the initial embedding and manually filtering out-of-concept or ambiguous concept terms…more details

Talk | Research Frontiers | Machine Learning | Advanced

To gain an edge in the markets quantitative hedge fund managers require automated processing to quickly extract actionable information from unstructured and increasingly non-traditional sources of data. The nature of these “alternative data” sources presents challenges that are comfortably addressed through machine learning techniques. We illustrate use of AI and ML techniques that help extract derived signals that have significant alpha or risk premium and lead to profitable trading strategies.

This session will cover the following topics:
The broad application of machine learning in finance
Extracting sentiment from textual data such as news stories and social media content using machine learning algorithms
Construction of scoring models and factors from complex data sets such as supply chain graph, options (implied volatility skew, term structure), Geolocational datasets and ESG (Environmental, Social and Governance)
Robust portfolio construction using multi-factor models by blending in factors derived from alternative data with the traditional factors such as fama-french five-factor model…more details

Workshop | Machine Learning | Data Visualization | Beginner – Intermediate

Working with mathematicians, data scientists, and domain experts at the University of Vermont Complex Systems Center, data visualization artist Jane Adams has developed strategies for prototyping exploratory graphs of high-dimensional data. In this 90-minute workshop, Adams shares some of these methods for data discovery and interaction, navigating a creative workflow from paper prototypes of visual hypotheses through web-based interactive slices, offering critical insight for clustering, interpolation, and feature engineering.

Talk

The goal of this talk is to demystify the application of artificial intelligence in the security industry. I will address common misconceptions and detail common use cases, while attempting to cut through the hype and inflated marketing claims for AI systems. I will walk through coding examples for training predictive models including spam detection and malware classification. In addition to discussing the benefits, I will also discuss potential pitfalls and challenges. The end of the talk will flip the thesis to discuss applications (or lack thereof) of cybersecurity in AI, detailing famous adversarial attacks on AI systems and methods to mitigate such attacks. Members of the target audience have a curiosity about how AI methodologies are applied in cybersecurity, but need not be experts…more details

Talk | Deep Learning | Machine Learning | Intermediate

Distributed representations of words has proven to be successful in addressing previous drawbacks of symbolic representations which treated words as atomic units of meaning. Symbolic representations treat words like an island, unable to capture similarity and relatedness information between. Word representations like Word2Vec, GLoVE, Fasttext on the other hand use distributional semantics and learn compressed representations of words by accounting contextual information. These representations are able learn meaningful analogical and lexical relationship which make them a popular choice in downstream NLP tasks…more details

Talk | Deep Learning | Beginner-Intermediate

In this talk, we’ll walk through the details and lessons learned from developing and deploying a visual search service at ShopRunner. You’ll learn to leverage Docker to develop code that is agnostic to CPU and GPU compute environments, evaluate your models (yes! even unsupervised ones), design a scalable service architecture with low latency and high availability, utilize your org’s continuous integration and continuous deployment (CI/CD) pipelines, and communicate what you’ve done with the rest of your org.

You’ll get the most out of this session if you have some familiarity with deep learning, and have deployed code within a production stack. Though we’ll discuss code snippets in Python, familiarity with any programming language will suffice.

Talk | Deep Learning | Machine Learning | Beginner – Intermediate

We will explore the use of modern deep learning approaches to object detection, semantic segmentation towards feature extraction and meaningful change detection in satellite imagery. We identify techniques towards sampling data, leveraging heterogeneous datasets, performance benchmarking and tuning models towards generalized performance. We will walk through the technical challenges involved with serving results from these deep learning models in a scalable manner along with essential metadata for interpretability and usability for our customers. We then explore approaches with human in the loop towards improved performance over time towards scalable location-based intelligence served across the entirety of the Earth’s landmass and discuss the challenges and opportunities at the frontiers of geospatial data and location-based insights. Specific applications of the Planet Analytics API towards geospatial intelligence will be discussed including data fusion from multiple geospatial data sources…more details

Talk | Machine Learning | AI for Engineers | Intermediate

The talk aims to introduce the attendees to the application of computer vision techniques to overhead imagery such as satellite, aerial and drone imagery. The emphasis is on object detection on satellite images as we share our learnings from dealing with those datasets (such as the xView Object Detection Challenge).

The attendees will learn about challenges common in such datasets, such as scale variance and images with more than three channels, as well as approaches to address them and the results of our experiments.

The session will also introduce datasets available in the growing field of CV on remote sensing data, as well as various use cases in disaster relief, refugee tracking, animal population counting, geospatial intelligence for business decisions and defense. The attendees will walk away with a good grasp of what is possible when machine learning at scale is applied to geospatial data…more details

Talk | Deep Learning | Machine Learning | Intermediate

With half of the world population online, and spending over 5 hours a day there, online communities are flourishing. It is now easier than ever for niche communities to form: gamers can find other players and form teams, dating adults can find better matches, students of particular subjects can find teachers and help each other. With faster networks images, audio, and video, are increasingly complimenting text, creating a richer experience.

However, as these communities grow wider and deeper, they can become a target for toxic behavior. Forums for underage users can be subverted with users attempting illicit solicitations and exploitation. Chat rooms can see participants engaging in cyberbullying and toxic language. According to Pew Research, over half of online users have seen offensive name calling and intentional embarrassment, while a quarter have witnessed physical threats and even prolonged harassment. This clearly has to stop. Businesses must now manage their communities in such a way that first and foremost protects their users, while also guarding their brand’s reputation…more details

Workshop | AI For Engineers | Open Source Data Science | Beginner – Intermediate

pomegranate is a Python package for probabilistic modeling that emphasizes both ease of use and speed. In keeping with the first emphasis, pomegranate has a simple sklearn-like API for training models and performing inference, and a convenient “lego API” that allows complex models to be specified out of simple components. In keeping with the second emphasis , the computationally intensive parts of pomegranate are written in efficient Cython code, all models support both multithreaded parallelism, out-of-core computations, and some models support GPU calculations. In this talk I will give an overview of the features in pomegranate, such as missing value support, demonstrate how the flexible provided by pomegranate can yield more accurate models, and draw examples from “popular culture” to inadvertantly prove how out of touch I am with today’s youth. I will also demonstrate how one can use the recently added custom distribution support to make neural probabilistic models, such as neural HMMs, using whatever your favorite neural network package is.

Talk | Deep Learning | Machine Learning | Intermediate

Neural networks dominate the modern machine learning landscape, but their training and performance still suffer from sensitivity to the empirical choice of training task hyperparameters. The aim of automated machine learning (AutoML) techniques is to automate and optimize the population of training tasks during the empirical parameter search, in order to maximise performance under limited computation resources. Current attempts for AutoML are focused on configurations common to deep learning models, such as architecture, loss function, learning rate and optimization algorithm. Nevertheless, the training data and its quality are considered constant, which is in contrast to their importance in determining the quality of the trained model. In this work, we propose an integrated approach to AutoML which includes, in addition to the off-the-shelf hyperparameter optimizer, parameterization over the metadata population selected for training.

Talk | Research Frontiers | Data Science for Good | All levels

We will discuss new techniques in Artificial Intelligence and Operations Research to help mitigate homelessness and how we are working with policy-makers to deploy and evaluate them in the open world. In particular, we discuss two problems and proposed algorithms: the first one is a preference elicitation algorithm that learns the moral priorities of policy-makers in terms of policy characteristics and outcomes (e.g., fairness by race, by gender); the second one is a data-driven optimization model and approximation algorithm to help design interpretable policies that meet the desiderata of policy-makers. We show that our proposed algorithms outperform the state of the art using real data from the Homeless Management Information System database. Throughout the talk, we discuss how the research approach that we take understands and anticipates the societal challenges arising from the deployment of AI technology in the open world (e.g., introduction of bias from use of standard ML, possibility that policy-makers give irrational/inconsistent answers)…more details

Workshop | Deep Learning | Machine Learning | Intermediate

Natural language processing is a key component in many data science systems that must understand or reason about text. Common use cases include question answering, paraphrasing or summarization, sentiment analysis, natural language BI, and entity extraction. This talk introduces the open-source Spark NLP library, which within two years have become the most widely used NLP library in the enterprise – by implementing state-of-the-art deep learning NLP research as a production-grade, fast and scalable library for Python, Java and Scala.

Spark NLP natively extends the Spark ML pipeline API’s which enabling zero-copy, distributed, combined NLP & ML pipelines, which leverage all of Spark’s built-in optimizations. Benchmarks and design best practices for building NLP, ML and DL pipelines on Spark will be shared. The library implements core NLP algorithms including lemmatization, part of speech tagging, dependency parsing, named entity recognition, spell checking and sentiment detection…more details

Talks | Research Frontiers | Machine Learning | Intermediate-Advanced

In this talk, Dr. Anju Kambadur, Head of AI Engineering at Bloomberg, will provide an overview of his group’s research activities over the past year across 5 key problem sets. He will then dive deeper into a recent paper he and his colleagues published at SIGIR ‘18 entitled “Weakly-Supervised Contextualization Of Knowledge Graph Facts,” which details a new neural fact contextualization method (NFCM) that can be used in knowledge graph fact contextualization tasks…more details

Talk | Machine Learning | AI for Engineers | Intermediate

Data scientists and machine learning engineers use a variety of open source projects in their everyday tasks: scikit-learn, SparkML, TensorFlow, Apache MXNet, Pytorch, etc. They make it very easy to get started, but as models become more complex and datasets become larger, training time and prediction latency become a significant concern. Here too, containers can help, especially when used with elastic on-demand compute services. In this session, we’ll show you how to scale machine learning workloads using containers on AWS (Deep Learning AMI and containers, ECS, EKS, SageMaker). We’ll discuss the pros and cons of these different services from a technical, operational and cost perspective. Of course, we’ll run some demos :)​

Tutorials | Data Visualization | Open Source | Beginner

In this talk we’ll explore what types to data the bootstrap has trouble with. Then we’ll discuss how to identify these problems in the wild and how to deal with the problematic data. We will explore simulated data and share the code to conduct the simulations yourself. However, this isn’t just a theoretical problem. We’ll also explore real Firefox data and discuss how Firefox’s data science team handles this data when analyzing experiments.

At the end of this session you’ll leave with a firm understanding of the bootstrap. Even better, you’ll understand how to spot potential issues in your data and avoid false confidence in your results…more details

Talk | Machine Learning | Research Frontiers | Intermediate

Representation learning, colloquially known as embeddings, has emerged as an important unifying theme in the machine learning community and is widely used in communities ranging from social media to computer vision and natural language processing. The core idea is to leverage large quantities of context-rich data, whether labeled or unlabeled, to ‘embed’ data points into vectors. These vectors can then serve as feature sets for classic machine learning classifiers like Logistic Regression. Embedding algorithms like word2vec and DeepWalk have yielded impressive results in natural language and graph processing pipelines. In the research community, there is a concerted effort now to build faster and better embedding algorithms for all kinds of heterogeneous datasets, including videos with tags and annotations, social media data and tables…more details

Workshops | DevOps & Managment | Beginner-Intermediate

As larger quantities of data are being stored and managed by enterprises of all kinds, NoSQL storage solutions are becoming more popular. Elasticsearch is a popular, high-performance NoSQL data storage option, but it is often unfamiliar to end users and difficult to navigate for day to day analytic tasks.

This presentation will briefly discuss the structure and benefits/drawbacks of Elasticsearch data storage, and describe in detail, with examples, how your end users can get data out of Elasticsearch data storage powerfully and with confidence. Attendees will be introduced to three packages designed for this work, elastic (R), elasticsearch-py (Python), and uptasticsearch (R and Python), and will see hands-on examples of how to use them…more details

Workshop | Deep Learning | Open Source Data Science | All Levels

This workshop shows how interpretability tools can give you not only more confidence in a model, but also help to improve model performance. Through this interactive workshop, you will learn how to better understand the models you build, along with the latest techniques and many tricks of the trade around interpretability. The workshop will largely focus on interpretability techniques, such as feature importance, partial dependence, and explanation approaches, such as LIME and Shap.
The workshop will demonstrate interpretability techniques with notebooks, some in R and some in Python. Along the way, workshop will consider issues like spurious correlation, random effects, multicollinearity, reproducibility, and other issues that may affect model interpretation and performance. To illustrate the points, the workshop will use easy to understand examples and references to open source tools to illustrate the techniques.

Workshop | DevOps & Managment | Machine Learning | Beginner-Intermediate

DevOps tools for getting code reliably to production have proven to be effective in the software engineering world. Today, ML Engineers are working at the intersection of data science and software engineering, and can leverage DevOps best practices to streamline their workflow and delivery process. This is what MLOps is all about.

At Manifold, we’ve developed processes to help ML engineers work as an an integrated part of your development and production teams, helping you to be deliberate, disciplined, and coordinated in your deployment process. In this workshop, Sourav and Alex will walk you through the some key learnings from using this Lean AI process…more details

Talks | Research Frontiers

To safely navigate roads and adapt to our ever-evolving cityscape, self-driving vehicles need a deep understanding of the world around us. We’ll explore how Waymo uses real-world and simulated data with deep learning to unlock new capabilities and build safer autonomous vehicles. We’ll also cover how Waymo uses data to develop machine learning at scale as it expands to new cities and geographies. As the only autonomous vehicle company that designs the full suite of self-driving hardware and software in-house, Waymo has developed an integrated system that allows each Waymo vehicle to process a large and diverse set of sensor data, ultimately allowing our vehicles to make informed, real-time decisions. We’ll touch on the benefits of designing and developing our hardware (LiDAR, radar, and cameras), software and manufacturing systems, how deep learning is used in various parts of self-driving from mapping, to perception, behavior prediction, and more, and what machine learning infrastructure is required to develop self-driving technology at scale…more details

Talk | Machine Learning | Intermediate

Next best action (NBA) is a technique that takes unique user history and characteristics into consideration and recommends the next actions that help the customer progressing towards business goals as quickly and smoothly as possible. It is not easy to design such a AI powered NBA engine. Ideally a hand-free NBA engine needs to handle the following problems. a) It should deal with incomplete historical feedback that are skewed towards a small set of actions; b) It should adapt to dynamic actions, which can be added or removed frequently due to seasonal changes or shifts in business strategies; c) It needs to optimize for multiple complex business objectives, which usually consist of reaching a set of target events or moving users to next more preferred stage; d) Most importantly, it has to learn and make decisions in realtime and at massive scale. Most of the solutions in the market only addresses a few of those challenges, due to model or technical challenges…more details

Talks | AI for Engineers | DevOps & Managment | Intermediate-Advanced

AI is introducing new frontiers in engineering and science. A fleet of cars can be monitored in real time and problems can be diagnosed before they happen. Smart medical devices and diagnostics can help save lives. However, there are many complexities associated with acquiring and managing engineering data, training predictive models, and deploying these models to be used in near real-time. This talk will discuss these complexities in the context of building a system to perform streaming predictions on sensor data…more details

Talk | Research Frontiers | Deep Learning | Intermediate

Content quality of product catalog plays a critical role in e-commerce. In particular, visual content such as product images influences customers’ engagement and purchase decisions more heavily than plain text. With the rapid growth of e-commerce and the advent of computer vision and AI, traditional content management systems are giving way to automated scalable systems. This talk presents such a machine learning driven system for evaluating, understanding and optimally selecting product images for customers. For a given product, the system aggregates images from various sources, analyzes and filters them for poor quality or non-compliant content, de-duplicates them, classifies them based on viewpoint and relevance to the product, and finally, selects a set of images with optimal count and quality. The system is empowered by an array of state-of-the-art technologies, ranging from deep image classification, deep regression, object detection to traditional computer vision…more details

Talks | Data Science Management | All Levels

Many businesses today are using traditional AI as an advanced modeling tool to create insights for critical business decision making. But insights alone are not enough. Evolutionary AI provides prescriptive guidance for your decision making to get you to optimized outcomes. Today, we can produce optimal solutions for the most complex multi-objective search-spaces by applying the creative and efficient capabilities of evolutionary computation. This session will show how such decision augmentation can be achieved via step by step hands-on examples, demos and interactive white-boarding to scope out several use cases of interest for the audience

Talk | DevOps & Managment | Open Source Data Science | Intermediate

One of the key bottlenecks in building machine learning systems is creating and managing the massive training datasets that today’s models learn from. In this talk, I will describe my work on data management systems that let users specify training datasets in higher-level, faster, and more flexible ways, leading to applications that can be built in hours or days, rather than months or years.

I will start by describing Snorkel, an open-source system for programmatically labeling training data that has been deployed by major technology companies, academic labs, and government agencies. In Snorkel, rather than hand-labeling training data, users write labeling functions which label data using heuristic strategies such as pattern matching, distant supervision, and other models…more details

Workshop | Machine Learning | Beginner

Nausea, vomiting, and diarrhea are words you would not frequently find in a natural language processing (NLP) project for tweets or product reviews. However, these words are common in healthcare. In fact, many clinical signs and patient symptoms (e.g. shortness of breath, fever, or chest pain) are only present in free-text notes and are not captured with structured numerical data. As a result, it is important for healthcare data scientists to be able to extract insight from unstructured clinical notes in electronic medical records.

In this hands-on workshop, the audience will have the opportunity to complete a Python NLP project with doctors’ discharge summaries to predict unplanned hospital readmission. The audience will learn how to prepare data for a machine learning project, preprocess text using a bag-of-words approach, train a few predictive models, evaluate the performance of the models, and strategize how to improve the models. The MIMIC III data set is used in this tutorial and requires requesting access in advance (an artificial dataset will be provided for those without access).

Tutorials | Data Science for Good | Machine Learning

Climate change is one of the greatest challenges facing humanity, and we, as machine learning experts, may wonder how we can help. Here we describe how machine learning can be a powerful tool in reducing greenhouse gas emissions and helping society adapt to a changing climate. From smart grids to disaster management, we identify high impact problems where existing gaps can be filled by machine learning, in collaboration with other fields. Our recommendations encompass exciting research questions as well as promising business opportunities. We call on the machine learning community to join the global effort against climate change…more details

Talk | Machine Learning | AI for Engineers | Intermediate

In this talk, Liang will introduce two projects in Airbnb Experiences, a new service of Airbnb, highlighting the use of causal inference in an E-Commerce search engine. A new service usually faces challenges brought by small data: inadequate data and noisy label information, which make it difficult to utilize user search logs. This problem can be easily solved when the user traffic is large enough to support large-scale randomized experiments. For an early-stage product, however, it is particularly difficult, if not impossible to launch a large number of A/B testing. In this talk, Liang will present how to solve this problem with causal inference. In particular, he will introduce basic concepts of Causal Inference, and present two case studies about improving the understanding of user behaviors, such as clicks and purchases, by utilizing causal analytics and designing natural experiments.

Talk | Open Source Data Science | Big Data | Intermediate

This talk reflects on the design and architecture of an effective modern big data platform that can ingest, store, and serve 100+ PB of data with minute level latency. We’ll walk you through the typical workflow of a data scientist or data analysts at Uber to explore data, discover desired datasets, access the data, run interactive queries, visualize the output, or prepare derived datasets for advanced analytics and machine learning use cases. The audience will leave the talk with greater insight into how things work in an extensible modern Big Data platform and will be inspired to re-envision their own data platform to make it more generic and flexible for their data scientists and analysts…more details

Workshops | Deep Learning | Intermediate

Deep Learning became known for beating human performance on image classification, and most applications and published examples for deep learning still focus on image processing. However, deep learning nowadays can also be applied successfully to time-series (or sensor) data and text. In this hands-on MATLAB workshop, you will see how easy it is to get started applying deep learning to various data types, including images, text, and time-series data. You’ll use an online MATLAB instance to perform the following tasks:
1. Train deep neural networks on GPUs in the cloud
2. Access and explore pretrained models
3. Build a CNN to solve an image classification problem
4. Use LSTM networks to solve a time-series and text analytics problem…more details

Workshop | Data Visualization | Open Source Data Science | Intermediate

In this hands-on workshop, we will use R to take public data from various sources and combine them to find statistically interesting patterns and display them in static and dynamic, web-ready maps. This session will cover topics including geojson and shapefiles, how to munge Census Bureau data, geocoding street addresses, transforming latitude and longitude to the containing polygon, and data visualization principles.

Participants will leave this workshop with a publication-quality data product and the skills to apply what they’ve learned to data in their field or area of interest…more details

Talk | Deep Learning | AI for Engineers | Beginner – Intermediate

Traditionally, NLP problems are complex tasks that requires large manually labelled datasets and complex ML models. Transfer learning and Deep Learning have made a breakthrough in Computer Vision, and a similar class of models are set to revolutionise NLP. Lars Hulstaert will discuss how models like ElMo, GPT-2 and UlmFit will reduce the entry barrier for extremely powerful NLP pipelines…more details

Talk | Intermediate

Much of the text classification problem has been treated as a supervised learning and requires large number of labeled data. However, this is not usually the case when it comes to real word use cases. The recent breakthroughs in NLP allow us to get use of pre-trained word embeddings from large corpora such as Google and Wikipedia. However, loading the pre-trained embeddings into memory via Gensim or Spacy is not computationally effective and it makes it harder to get run time word vectors. Given this circumstances and lack of labeled data, I propose to use Google and Elmo word embeddings over a new Python package called “Magnitude” without loading the entire corpora into memory. Recently, I built a model for a knowledge management platform to suggest topics for the entries given by the end users. There was no labeled data, and the users start with zero entries but assign tags (topics) one by one while the system calculates topic representation vectors for four different approaches…more details

Talk | Research Frontiers | Deep Learning | Intermediate – Advanced

Deep neural networks now enable machines to learn to solve problems that were previously been easy for humans but difficult for computers, like playing Atari games or identifying lions and jaguars in photos. But how do these neural nets actually work? What concepts do they learn en route to their goals? We built and trained the networks, so on the surface these questions might seem trivial to answer. However, network training dynamics, internal representations, and mechanisms of computation turn out to be surprisingly tricky to study and understand, because networks have so many connections — often millions or more — that the resulting computation is fundamentally complex.

This high fundamental complexity enables the models to master their tasks, but we find now that we need something like neuroscience just to understand the AI models that we’ve constructed!..more details

Talk | Deep Learning | AI for Engineers | Intermediate – Advanced

A portion of gross merchandise volume in e-commerce is driven by categories such as fashion, home, furniture, lifestyle and apparel. These categories differ from other categories in terms of need for discovery and the way people making purchase decisions. They require visual appeal and visual discovery where people want to navigate them through how similar they are in terms of visual appearance rather than any other attribute of the products.

Most of the search engines and e-commerce search engines provide only text based searches and it utilizes textual attributes of products. This does not provide a good search and recommendations experience for categories that are rich in terms of visual information…more details

Talk | Machine Learning | Beginner – Intermediate

Class imbalance is extremely common in real-world datasets. Yet so many data scientists use machine learning algorithms without considering this challenge. The presence of class imbalance violates key assumptions of many algorithms, which leads to flawed models. This talk will:
– Explain why such imbalances occur, drawing on research into the theory of class imbalance, class overlap, and noise.
– Describe error metrics that can expose or mitigate class imbalance (including ROC curves, precision, and recall).
– Discuss some solutions including pre-processing techniques (over- and under-sampling, weighting), modifications to existing algorithms, and post-processing techniques (threshold selection, cost-based classification).

You’ll leave this presentation with the ability to recognize class imbalance and the confidence to overcome it.

Workshop | Machine Learning | Intermediate

In this workshop we will dive deep into what it takes to build and deliver an always-on “real-ish time” predictive analytics pipeline with Spark Structured Streaming.

The core focus of the workshop material will be on how to solve a common complex problem in which we have no labeled data in an unbounded timeseries dataset and need to understand the substructure of said chaos in order to apply common supervised and statistical modeling techniques to our data in a streaming fashion.

The example problem for the workshop will come from the telecommunications space but the skills you will leave with can be applied to almost any domain as long as you sprinkle in a little creativity and inject a bit of domain knowledge…more details

Workshop | Machine Learning | Data Visualization | Intermediate

With the recent popularity of machine learning algorithms such as neural networks and ensemble methods, etc., machine learning models become more like a ‘black box’, harder to understand and interpret. To gain the end user’s trust, there is a strong need to develop tools and methodologies to help the user to understand and explain how predictions are made. Data scientists also need to have the necessary insights to learn how the model can be improved. Much research has gone into model interpretability and recently several open sources tools, including LIME, SHAP, and GAMs, etc., have been published on GitHub. In this talk, we present Microsoft’s brand new Machine Learning Interpretability toolkit which incorporates the cutting-edge technologies developed by Microsoft and leverages proven third-party libraries. It creates a common API and data structure across the integrated libraries and integrates Azure Machine Learning services. Using this toolkit, data scientists can explain machine learning models using state-of-art technologies in an easy-to-use and scalable fashion at training and inferencing time.