ODSC East 2020

Training | Kickstarter | Machine Learning | Beginner

Machine learning has become an indispensable tool across many areas of research and commercial applications. From text-to-speech for your phone to detect the Higgs boson, machine learning excels 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 formulating a problem as a supervised machine learning problem, creating the necessary training data and applying and evaluating a machine learning algorithm. This workshop should give you all the necessary background to start using machine learning yourself…more details

Training | Machine Learning | ML for Programmers | Intermediate

scikit-learn is a machine learning library in Python, that has become a valuable tool for many data science practitioners. This workshop will go beyond the basics and show how to effectively evaluate and tune algorithms. We will also discuss the most important machine learning algorithms that you’re likely to see in practice, how and when to use them, and some details about how they work internally. The session will focus on linear models for classification and regression and tree-based models, including random forests…more details

Training | Machine Learning | ML for Programmers | Advanced

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 and advanced model evaluation. Model evaluation is an underappreciated aspect of machine learning, but using the right metric to measure success is critical. Practitioners are often faced with imbalanced classification tasks, where accuracy can be uninformative or misleading. We will discuss other metrics, when to use them, and how to compute them with scikit-learn. We will also look into how to build processing pipelines using scikit-learn, to chain multiple preprocessing techniques together with supervised models, and how to tune complex pipelines…more details

Training | Machine Learning | ML for Programmers | Advanced

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 and how to build your own models or feature extraction methods that are compatible with scikit-learn. We will also discuss different approaches to feature selection and resampling methods for imbalanced data. Finally, we’ll discuss how to do the classification of text data using the bag-of-words model and its variants…more details

Full-Day Training | Kick-starter | Open-source | Intermediate

Whether in R, MATLAB, Stata, or Python, modern data analysis, for many researchers, requires some kind of programming. The preponderance of tools and specialized languages for data analysis suggests that general purpose programming languages like C and Java do not readily address the needs of data scientists; something more is needed.

In this training, you will learn how to accelerate your data analyses using the Python language and Pandas, a library specifically designed for interactive data analysis. Pandas is a massive library, so we will focus on its core functionality, specifically, loading, filtering, grouping, and transforming data. Having completed this workshop, you will understand the fundamentals of Pandas, be aware of common pitfalls, and be ready to perform your own analyses…more details

Training | Machine Learning | Intermediate

An ML application in production must address all of the issues of modern software development methodology, as well as issues unique to ML and data science. Often ML applications are developed using tools and systems which suffer from inherent limitations in testability, scalability across clusters, training/serving skew, and the modularity and reusability of components. In addition, ML application measurement often emphasizes top level metrics, leading to issues in model fairness as well as predictive performance across user segments.

We discuss the use of ML pipeline architectures for implementing production ML applications, and in particular we review Google’s experience with TensorFlow Extended (TFX), as well as the advantages of containerizing pipeline architectures using platforms such as Kubeflow. Google uses TFX for large scale ML applications, and offers an open-source version to the community. TFX scales to very large training sets and very high request volumes, and enables strong software methodology including testability, hot versioning, and deep performance analysis…more details

Training | Open-source | Kick-starter | Beginner

Recommendation systems are fundamental to essentially every business on the planet. From providing the shows and movies you should watch, the books and articles you should read, the products you should purchase, or the people you should date making good recommendations can make or break your business. This workshop will give a general introduction to recommendation systems, as well as introduce practical tools for you to apply recommendation systems to your own use cases. Though the primary focus of this workshop will be to understand the different methods you might use to make recommendations, we will also touch briefly on common difficulties with building effective recommendation systems, how to deploy your recommendations, and how to evaluate whether your recommendations are effective…more details

Training | Machine Learning | MLOps & Data Engineering | Beginner-Intermediate

In this half-day hands-on training, we will use free-tier resources in the Google Cloud Platform (GCP) to introduce learners to the practical use of cloud computing resources in data science and machine learning. Learners should have some experience with data analytics, data science or machine learning. Learners should also have a Gmail account with no former GCP use associated with it, or be willing to create such an account. While fluency in R or Python will be very helpful, it is not rigorously required, as well-annotated scripts will be provided. No previous exposure to or use of cloud computing is required; this is introductory-level in terms of its cloud computing assumptions. This training will be useful for those considering cloud adoption, interested in data engineering, or interested in working with public data as citizen scientists…more details

Training | Data Visualization | Machine Learning | Beginner-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 3.5 hr 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.

Upon completing this tutorial, you will be:
– familiar with how to use the NetworkX and nxviz Python packages for modeling and rationally visualizing networks,
– able to load node and edge data from a Pandas data frame,
– familiar with object-oriented and matrix-oriented representations of graphs,
able to find paths between nodes, interesting structures in graphs, and projections of bipartite graphs.
– (if time permits) able to use matrix operations to simulate the diffusion of information on networks…more details

Full-Day Training | Deep Learning | Machine Learning | Beginner-Intermediate

In this workshop we will explore Reinforcement Learning, starting from its fundamentals and ending creating our own algorithms. We will use OpenAI gym to try our RL algorithms. OpenAI is a non profit organization that want committed to open source all their research on Artificial Intelligence. To foster innovation OpenAI created a virtual environment, OpenAi gym, where it’s easy to test Reinforcement Learning algorithms. We then will also explore other RL frameworks and more complex concepts like Policy gradients methods and Deep Reinforcement learning, which recently changed the field of Reinforcement Learning. In particular, we will see Actor-Critic models and Proximal Policy Optimizations that allowed OpenAI to beat some of the best Dota players…more details

Traning | Data Vizualization | ML for Programmers | Beginner-Intermediate

This workshop will take one through the steps associated with an end-to-end machine learning campaign: data retrieval; data curation; model construction, evaluation, selection and interpretation; 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.

The workshop will have three main foci:
1. infrastructure: instantiating the computational environment; loading packages; loading data
2. computation: data curation, transformation, and analysis; model construction, evaluation, and interpretation
3. communication: creating tables, charts, and graphs; weaving all components into data narrative…more details

Training | Machine Learning | MLOps & Data Engineering | Intermediate-Advanced

In this workshop, you will learn the best techniques that can be applied manually or automatically to validate and monitor statistical models.
Techniques below will be discussed and demonstrated to perform a full model validation:
Techniques used for initial validation:
Bias and variance
Model selection
Hyperparameters selection
Model interpretation
Adversarial validation o
Techniques used for model monitoring
Bias and variance consistence
Model degradation
Model interpretation consistence…more details

Training | Open-source | Kick-starter | Beginner

Data Science and Machine learning have been synonymous with languages like Python. Libraries like Numpy and Pandas have become the de facto standard when working with data. The DataFrame object provided by Pandas gives us the ability to work with heterogeneous unstructured data that is commonly used in “real world” data. New learners are often drawn to Python and Pandas because of all the different and exciting types of models and insights the language can do and provide but are awestruck when faced with the initial learning curve. By the end of this tutorial, you should have a solid foundation in working with datasets in Python. The last topic of encoding dummy variables segues into using other libraries, such as scikit-learn and stats models to fit models on your data…more details

Bootcamp | Kickstarter | Open-source | Beginner

By completing this workshop, you will develop an understanding of relational models of data, how SQL is used to retrieve that data, and how to join tables, aggregate information, and answer data science questions. You will also become familiar with many of the common types of SQL databases, how to access information in a database from the command line, and how to integrate database access from within Python.

Lesson 1: Relational Databases and Foundational SQL
Familiarize yourself with relational databases and the SQL syntax necessary to retrieve information from tables in a database. At the end of this lesson, you will be able to comfortably explore a database and retrieve filter, and sort information from a table…more details

Training | MLOps & Data Engineering | Intermediate

Ray is an open-source distributed framework from U.C. Berkeley’s RISELab that easily scales Python applications from a laptop to a cluster, with an emphasis on ML/AI systems, such as reinforcement learning. It is now used in many production deployments. In this tutorial, we’ll use several hands-on examples to explore the problems that Ray solves and the useful features it provides, such as rapid distribution, scheduling, and execution of “tasks” and management of distributed stateful “serverless” computing. We’ll see how it’s used in several ML libraries (and play with examples using those libraries). You’ll learn when to use Ray and how to use it in your projects…more details

Training | Deep Learning | Intermediate-Advanced

In this session, we will focus on the application and practice of building a multi-layer neural network in Python code. We will walk through a lightweight neural network framework and discuss how the concepts from the morning’s session were reduced to code, including optimization, regularizerization, and backpropagation. We will step through a concrete example of building an autoencoder for image compression. Participants will have the option to download and run all of the neural network code on their own computers. No GPU required. Participants are also welcome to observe and absorb…more details

Training | Deep Learning | Kick-starter | Beginner

This session will focus on the foundational concepts of neural networks and deep learning. Concrete examples will be used to illustrate abstract concepts and methods. Participating in this workshop you will gain an intuition for the principles underlying modern neural networks including multilayer perceptrons, autoencoders, convolutional neural networks, and recurrent neural networks including long short-term memory. No prior experience with neural networks or machine learning required…more details

Training | Machine Learning | Open-source | Beginner-Intermediate

The rapid progress in machine learning (ML) and the massive increase in the availability and diversity of data has enabled novel approaches to quantitative investment. It has also increased the demand for the application of data science to develop both discretionary and algorithmic trading strategies.
In this workshop, we will cover popular use cases for ML in the investment industry, and how data science and ML fit into the workflow of developing a trading and investment strategy from the identification and combination of alpha factors to strategy backtesting and asset allocation.
The workshop uses Python and various standard data science and machine learning libraries like pandas, scikit-learn, gensim, spaCy as well as TensorFlow and Keras. The code examples will be presented using jupyter notebooks and are based on my book ‘Machine Learning for Algorithmic Trading’…more details

Training | MLOps & Data Engineering | ML for Programmers | Beginner-Intermediate

By attending this workshop, you will learn how you can become a contributor to Apache Airflow project. You will learn how to setup development environment, how to pick your first issue, how to communicate effectively within the community and how to make your first PR – experienced committers of Apache Airflow project will give you step-by-step instructions and will guide you in the process. When you finish the workshop you will be equipped with everything that is needed to make further contributions to the Apache Airflow project.

Lesson 1: Development environment

Learn how to setup and configure Apache Airflow development environment in less than 10 minutes. You will learn how modern tooling is used to provide consistent, reproducible environment for making contributions for distributed contributors of Apache Airflow. You will learn how to run tests and debug the code, and make sure that the tests are passing in the Continuous Integration environment as well as locally…more details

Training | MLOps & Data Engineering | Machine Learning | Beginner-Intermediate

ML development brings many new complexities beyond the traditional software development lifecycle. Unlike in traditional software development, ML developers want to try multiple algorithms, tools, and parameters to get the best results, and they need to track this information to reproduce work. In addition, developers need to use many distinct systems to productionize models. To solve these challenges, MLflow, an open-source project, simplifies the entire ML lifecycle. MLflow introduces simple abstractions to package reproducible projects, track results, and encapsulate models that can be used with many existing tools, accelerating the ML lifecycle for organizations of any size…more details

Training | Deep Learning | Machine Learning | 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, 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 | MLOps & Data Engineering | Beginner-Intermediate

Machine learning is usually taught from tutorials using small, clean datasets put into data-frames and orchestrated with Jupyter notebooks; all done in one, in-memory, local environment. This is a fine style for presenting a new topic and teaching the main ideas, but unfortunately, these patterns are not conducive to the delivery of real production applications at scale. In this workshop, we show how to maintain data science productivity as well as collaborate effectively and deliver value continuously and seamlessly. We demonstrate and guide the participants through CI/CD practices for machine learning and a new pattern of working that avoids most of the pitfalls of the typical approach.

Participants will learn how to utilize new patterns of repeatable continuous model development to collaborate effectively and deliver value continuously and seamlessly in industrial data science projects using Continuous Integration (CI) and Continuous Delivery (CD) practices….more details

Training | Machine Learning | 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, and then I will show how each one can be adapted to a novel use case that may be less obvious, perhaps producing significantly surprising results in some other domain. Plus, I will suggest some new opportunities that may come from interesting combinations of data and algorithms. The point of these exercises is to demonstrate how data scientists can create even more value, beyond that which is expected, from our data assets and our algorithmic talents…more details

Training | Machine Learning | Intermediate

We will present at least 10 examples and suggested solutions of cold-start problems (i.e., that move from a bad initial random guess to a good, perhaps optimal, solution), covering a variety of different algorithms and applications, focused primarily on unsupervised learning, but with some supervised learning examples also. We will also introduce related concepts and their importance, including the objective function, genetic algorithms, backpropagation, gradient descent, and meta-learning. Those concepts represent the true keys that unlock performance in a cold-start challenge. Those are the magic ingredients in most of the examples that we will present. At the end of the workshop, you should be empowered, enabled, and emboldened to tackle similar machine learning challenges problems in other domains. After all, data are data, math is math, and good experience is transferable!..more details

Training | Computer Vision | Deep Learning | Intermediate

In this tutorial, we will introduce how to build an object detection model. Specifically, we will build an object detection model that identifies chess pieces (a custom dataset provided by the presenter). In doing so, participants will gain insight into the fundamentals of computer vision: structuring a good problem for object detection, dataset collection and annotation, data preparation through preprocessing, data augmentation to support a well-fit model, training a model, debugging a model’s fit, and using the model for inference.

We will introduce Keras and TensorFlow as our specific libraries for writing computer vision models and train a Yolov3 (You Only Look Once) model for real-time detection. Participants will leverage Colab for GPU compute.

Session Outline
– Introduction to Computer Computer Vision Problems
– Data Preparation
– Data Preprocessing and Augmentation
– Training Our Yolov3 Model
– Inference and Q&A…more details

Full-Day Training | MLOps & Data Engineering | Machine Learning | Intermediate-Advanced

Machine Learning and Graph Processing (e.g., Knowledge Graphs) have been two of the main trends over the past years.
Many powerful Machine Learning algorithms are based on graphs, e.g., Page Rank (Pregel), Recommendation Engines (collaborative filtering), text summarization and other NLP tasks. There are even more applications once we consider data pre-processing and feature engineering which are both vital tasks in Machine Learning Pipelines. In this tutorial, we will consider the symbiosis of graphs and Machine Learning. We will start by looking at popular graph algorithm and their value for feature engineering for Machine Learning models. We know about the value of high quality and quantity for building high-quality machine learning models, but for operating a production-grade machine learning pipeline Metadata is equally important. In this part, we will look at leveraging graphs to capture metadata and provenance information of our machine learning ecosystem…more details

Training | Deep Learning | Beginner

Over the past couple of years, PyTorch has been increasing in popularity in the Deep Learning community. What was initially a tool used by Deep Learning researchers has been making headway in industry settings. 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. Furthermore, a link to a clean documented GitHub repo with the solutions of the examples covered will be provided…more details

Training | Data Visualization | Kick-starter | All Levels

In this complete hands-on training session, you will learn to turn your data into interactive dashboards, how to create stories with data and share these dashboards with your audience. We will begin with a quick refresher of basics about design and information literacy and discussions about practices for creating charts and storytelling utilizing best visual practices. Whether your goal is to explain an insight or let your audience explore data insights, using Tableau’s simple drag-and-drop user interface makes the task easy and enjoyable.
In this session, we’ll cover intermediate and advanced tableau functionality:
1. Database connectivity
2. Blending and joins using multiple data sources, web connectors and plugins,
3. Perform data analyses and create graphs from a real-world dataset, using Tableau Public (free to use)
4. Deeper Analysis – Trends, Clustering, Distributions, and Forecasting
5. Table Calculations, Sets, Filters, Level of Detail expressions, Parameters
6. Spatial analytics – Using Maps and Geocoding
7. Right and Wrong way to build Dashboards and Best Practices…more details 

Keynote

When it comes to model development and deployment, organizations should have the freedom to choose different languages, tools, techniques and environments because a modeling melting pot can fuel innovation and creativity. But without a structured process to coordinate all the different pieces, model management can feel more like the wild, wild west – no governance, no oversight, and no strategy for putting analytic insights to work. This is where SAS® Open Model Manager can help. This new offering from SAS can wrangle the model management process by enabling IT and analytics professionals to register, deploy, and monitor open source models in a consistent, repeatable manner. Users will be able to compare and assess their models, manage champion and challenger models, and access built-in performance reports to quickly evaluate whether to retrain, retire or develop new ones. SAS Open Model Manager integrates with Python and R and is intended for organizations that use open source tools to build models and need model management capabilities…more details

Workshop | NLP | Deep Learning | Intermediate-Advanced

The speaker will review case studies from real-world projects that built AI systems using Natural Language Processing (NLP) in healthcare. These case studies cover projects that deployed automated patient risk prediction, automated diagnosis, clinical guidelines, and revenue cycle optimization. He will also cover why and how NLP was used, what deep learning models and libraries were used, and what was achieved. Key takeaways for attendees will include important considerations for NLP projects including how to build domain-specific healthcare models and using NLP as part of larger and scalable machine learning and deep learning pipelines in a distributed environment…more details

Workshop | ML for Programmers | Deep Learning | Intermediate

Powerful Machine Learning models trained using various frameworks such as scikit-learn, PyTorch, TensorFlow, Keras, and others can often be challenging to deploy, maintain, and performantly operationalize for latency-sensitive customer scenarios. Using the standard Open Neural Network Exchange (ONNX) model format and the open source cross-platform ONNX Runtime inference engine, these models can be scalably deployed to cloud solutions on Azure as well as local devices ranging from Windows, Mac, and Linux to various IoT hardware. Once converted to the interoperable ONNX format, the same model can be served using the cross-platform ONNX Runtime inference engine across a wide variety of technology stacks to provide maximum flexibility and reduce deployment friction.

In this workshop, we will demonstrate the versatility and power of ONNX and ONNX Runtime by converting a traditional ML scikit-learnpipeline to ONNX, followed by exporting a PyTorch-trained Deep Neural Network model to ONNX. These models will then be deployed to Azure as a cloud service using Azure Machine Learning services, and to Windows or Mac devices for on-device inferencing…more details

Workshop | Machine Learning | Open-source | Intermediate

Gradient boosting is a powerful machine-learning technique that achieves state-of-the-art results in a variety of practical tasks. For a number of years, it has remained the primary method for learning problems with heterogeneous features, noisy data, and complex dependencies: web search, recommendation systems, weather forecasting, and many others.

Some problems contain different types of data, including numerical, categorical and text data. In this case, the best solution is either building new numerical features instead of text and categories and pass it to gradient boosting, or using out-of-the-box solutions for that.

CatBoost is the first Gradient Boosting library to have text features support out-of-the-box. This talk will walk you through the main features of this library including the way it works with texts.

This workshop will feature a comprehensive tutorial on using CatBoost library.

We will walk you through all the steps of building a good predictive model for data with text and numerical features.

Workshop | Deep Learning | MLOps & Data Engineering | Advanced

In this tutorial, we will guide you through hands-on examples of integrating tf.keras model with different data input sources through tf.data in production environments, from simple csv/json files to SQL databases and to cloud data warehouse services such as Google Cloud BigQuery. We will also cover Apache Kafka as a data input to illustrate the streaming data pipeline architecture for continuous data processing with machine learning. As a bonus, attendees will learn the basics of distributed machine learning and its production usage…more details

Workshop | ML for Programmers | Open-source | Beginner-Intermediate

From a consumption perspective, we discuss how to work with the httr, curl, and jsonlite packages. We show examples of constructing API requests, submitting them, and working with the responses. We discuss tips and tricks to optimize the workflow.
to interface with web REST APIs for many purposes.

From a command & control perspective, we discuss several packages included in the cloudyr (http://cloudyr.github.io/) project, which allow you to manage and execute tasks in the cloud. We show examples of creating resources and augmenting your data using APIs.

Last, we discuss the plumber R package, which allows users to expose existing R code as a service available to others on the Web. We discuss how to take models and functions and expose them as APIs so others can use them…more details

Training | Open-source | Intermediate

This tutorial will introduce you to the wonderful world of Bayesian data science through the lens of probabilistic programming in Python. In the first half of the tutorial, we will introduce the key concepts of probability distributions via hacker statistics, hands-on simulation, and telling stories of the data-generation processes. We will also cover the basics of joint and conditional probability, Bayes’ rule, and Bayesian inference, all through hands-on coding and real-world examples. In the second half of the tutorial, we will use a series of models to build your familiarity with PyMC3, showcasing how to perform the foundational inference tasks of parameter estimation, group comparison (for example, A/B tests and hypothesis testing), and arbitrary curve regression. By the end of this tutorial, you will be equipped with a solid grounding in Bayesian inference, able to write arbitrary models, and have experienced basic model checking workflow…more details

Workshop | MLOps & Data Engineering | Machine Learning | Beginner-Intermediate

The speed in which intelligent decision making can take place is changing the very fabric of the modern industry. From autonomous routing decisions for real-time traffic avoidance to intelligent integrated systems for customer service and assisted analysis or forecasting, the time to correct decision can be a huge differentiator. During the course of this workshop, you will learn how to harness the power of Apache Spark to build up a data ingestion pipeline that can process and learn from streaming data. These learnings can then be applied to make simple decisions using SparkSQL in a parallel streaming system…more details

Workshop | Machine Learning | Open Source | Intermediate

In this workshop, we will work hands-on using XGBoost with real-world data sets to demonstrate how to approach data sets with the twin goals of prediction and understanding in a manner such that improvements in one area yield improvements in the other. Using modern tooling such as Individual Conditional Expectation (ICE) plots and SHAP, as well as a sense of curiosity, we will extract powerful insights that could not be gained from simpler methods. In particular, attention will be placed on how to approach a data set with the goal of understanding as well as prediction…more details

Workshop | ML for Programmers | MLOps & Data Engineering | Intermediate-Advanced

Organizations have long used relational databases for a wide variety of data-intensive applications. Data scientists and analysts need to understand how to work with relational databases, particularly for common data science tasks, such as finding, exploring, analyzing and extracting data within a relational database. SQL is an expressive declarative language designed for working with tabular data. Since its inception, new features have been added that allow for complex expressions and computation, such as regular expressions, sliding window functions, and analytical operations that create non-relational results. Understanding the advanced features of SQL will help data scientists reduce the need for programming custom data manipulation functionality…more details

Workshop | Data Visualization | Beginner-Intermediate

Access to high-quality information to make good decisions is necessary for society to function. Good data level the playing field for businesses and individuals alike. It is necessary at every level of economy and society—to help small businesses succeed, schools serve parents and students, and central banks make sound policy. Those data must be an accurate reflection of the people in our country.
We must design a new system that will produce public data that are useful at all levels—and make scientific, careful, and responsible use of many newly available data, such as administrative records from agencies that administer government programs, data generated from the digital lives of citizens, and even data generated within the private sector. This presentation will paint a picture of what this new system could look like, focusing on the innovations necessary to change the existing system, with the goal of providing useful and timely data from trusted sources so that we, the people, have the information necessary to make better decisions.
The presentation particularly focuses on the role of data science and computer science in driving data innovation for evidence based policy…more details

Workshop | Machine Learning

Target leakage is one of the most difficult problems in developing real-world models. It occurs when training data gets contaminated with information that will not be known at prediction time. Data collection, feature engineering, partitioning, and model validation are all potential sources of data leakage. This talk offers real-life examples of data leakage at different stages of data science projects, discusses countermeasures, and lays out best practices for model validation…more details

Workshop | MLOps & Data Engineering | Intermediate

Kedro is a development workflow tool open sourced by QuantumBlack, a McKinsey company. Many data science teams have started using the library for their pipelines but are unsure how to integrate with other model tracking tools, such as MLflow. In this tutorial, we will give an overview of Kedro and MLflow and demo how to leverage the best of both. The goal of this session is to demonstrate how Kedro and MLflow fit together in a scalable AI architecture. To start, we will give an overview of Kedro and an overview of MLflow: – What are they used for? – What functionality do they provide? – How do they compare as tools? Next, we will walk through a demo of a Kedro project that has MLflow integrated into it. Finally, we will go over deployment options…more details

Workshop | Deep Learning | Beginner-Intermediate

As of today, PyTorch has become the dominant tool used in machine learning research. It comes bundled with a ton of features, but while the basics are often well explored in many tutorials, some more advanced functionality and workflows might deserve more attention than they usually get. In this tutorial, I’ll be aiming to improve the foundations and understanding of the basics of PyTorch, while also highlighting the parts that are often overlooked or misunderstood. Note that I will assume knowledge of some basic abstractions, but those should not exceed what can be found in the official “Getting started” tutorial…more details

Tutorial | Research Frontiers | Intermediate-Advanced

Thanks to advances in imitation and reinforcement learning techniques, we can now train intelligent agents to accomplish a diverse range of goals. But if we want to create household robots or personal assistants that can take advantage of this diversity, we need to give users some way to tell them what to do! This tutorial will focus on humans’ favorite tools for communicating goals and plans: natural language. We’ll assume basic familiarity with supervised learning and RL, and begin with a review of core machine learning techniques useful for natural language instruction following problems. The body of the talk will focus on modeling techniques for instruction following problems in different kinds of environments and data conditions. We’ll conclude with a survey of other applications for the tools we’ve built, including instruction generation, interpretability, and machine teaching…more details

Workshop | Computer Vision | Deep Learning | Beginner-Intermediate

This workshop will begin with an overview of common real-world tasks in the CV domain, including examples of problems our customers have faced in recent years. We will then give a brief introduction to deep learning models for CV. The main part of this session will demonstrate how to train and evaluate CV models by executing notebooks based on PyTorch’s Fast.ai and Torchvision libraries. We will start with image classification, how to fine-tune a pre-trained ImageNet model on a custom dataset, and show how to deploy the model to the cloud. Next, we will train an object detection model and extend the model to segmentation masks and keypoints. Finally, we will build an image similarity system and demo a fast image retrieval solution that can handle large amounts of images. We conclude by providing links to all examples and material for the audience to continue their learning journey…more details

Tutorial | Machine Learning | Intermediate

Most implementations of supervised machine learning algorithms are designed to work with complete datasets but datasets are rarely complete. I will quickly review the three types of missing data (missing completely at random, missing at random, missing not at random) and a couple of simple but often misleading ways to impute. I will spend most of the time describing three advanced methods for handling missing data: multiple imputation, the reduced-feature (aka pattern submodel) approach, and XGBoost, which is one of the few machine learning algorithms that works with incomplete datasets. We will discuss the advantages and limitations of each of these methods. By the end of the workshop, you will have a deeper understanding of the intricate nature of missing data and you will have multiple state of the art techniques in your arsenal to deal with them. I will use python to implement and visualize the techniques, relying on packages like pandas, scikit-learn, xgboost, and matplotlib or plotly for visualizations…more details

Workshop | ML for Programmers | Kick-starter | Beginner-Intermediate

When building a data science team it is important to document and record work flows in both R and python. A data science department can use installable git controlled packages to form a foundation of code and best practices. In addition, libraries dramatically cut down the time and effort required for a team to bring work to production. Additionally, package development is the first step for engineers and scientists aiming to contribute their unique ideas back to the community.

This workshop aims to teach the basics of package development in both R and python in 90 minutes. We will touch upon why a data science team should strive to be fully fluent in both languages. We will show simple R package development and simple Python package development. Finally, we will demonstrate how one can use an open source package to test the interface similarity between R and Python packages designed to support identical workflows…more details

Tutorial | Open-source | Machine Learning | Intermediate-Advanced

Data science has many powerful tools for using observational data to create models that solve prediction and classification problems. Many of our real world questions are causal, however, and it is well known that predictive models do not answer causal questions. A common attitude is that only controlled experiments can be used to produce causal information, but there are rich fields of statistics and machine learning covering correct methods for doing causal inference and causal discovery from observational data alone. Such methods are increasingly being used in practice, as experimentation is often costly, slow, unethical, or impossible. I will survey current methods for using observational data to answer causal questions, briefly review the theory and assumptions behind those methods, and demonstrate their use with open source software packages…more details

Talk | Research Frontiers | Machine Learning | Beginner-Intermediate

Recent advances in Artificial Intelligence (AI) and Machine Learning (ML) have relied on access to modern computing hardware, massive quantities of data, and advanced algorithms that leverage high-performance computing centers such as the MIT Lincoln Laboratory Supercomputing Center (LLSC). While AI will play in important role in many organizations, rapid adoption of AI is often limited by legacy hardware, siloed and messy data, and sparse/unsupervised ML algorithms in data starved application domains. As a world-leader in developing high-performance computing (HPC) tools that are easy-to-use without compromising performance, the LLSC has been developing a number of novel technologies that aim to overcome these technical hurdles that restrict easy adoption of AI. This research in “Fast AI” is pillared on modern computing, big data management, and interfaces & algorithms. In this talk, I will discuss the AI landscape; highlight a few AI adoption challenges; provide an overview of research that simplifies AI adoption; and discuss novel AI applications to cybersecurity and video recognition…more details

Talk | Data for Good | Machine Learning | Beginner-Intermediate

Two concurrent phenomena, digitization of cities and urbanization of the Internet, create new opportunities and complexities for cities. Although previous theory conceptualized cities as complex social-ecological-technical systems-of-systems with data as intermediate layers, ongoing ad-hoc smart city development has shaped a fragmented data landscape with organizational barriers and socio-technical conflicts. Large volume, variety, and velocity of data bring methodological and practical challenges for analyzing holistic urban systems with real-world data. Focusing on New York City, this talk discusses cross-domain data mining, integration, and modeling. In particular, it targets on three sub-problems in (1) data integration for quantifying hyper-local urban condition; (2) multivariate models for analyzing urban phenomena driven by cross-domain factors; and (3) unstructured data mining and knowledge discovery for information integration across multiple cities. After an overview of the research context and current methods, three research projects investigated methods for integrated analytics at hyper-local, micro, and city scale, with implementations related to urban built-ecological-socioeconomic systems. As the cities becoming increasingly smart and connected, an integrated and systematic data intelligence framework becomes critical for supporting better-informed, data-supported and collaborative urban systems…more details

Talk | Machine Learning | ML for Programmers | 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 | Machine Learning | Data Visualization | Beginner-Intermediate

I’m planning the family summer vacation. This year is going to be a little different. This year I’m going to use Python, Machine Learning and Oracle Cloud to plan the vacation. It will be a driving holiday in Scotland, but what places do we visit and in what order? Using details of local landmarks and family personal interest, Python and machine learning are used to identify different groupings of local landmarks matching those interests. Combining the outputs of the Python Machine Learning a driving route will be planned to optimise the number of landmarks to visit, where to stay and what else to visit and then saving the details to Oracle Cloud. Do one thing for me! Don’t tell my family this is how our vacation is being planned…more details

Talk | ML for Programmers | Beginner

In the last ten years, there have been a number of advancements in the study of Hamiltonian Monte Carlo algorithms that have enabled effective Bayesian statistical computation for much more complicated models than were previously feasible. These algorithmic advancements have been accompanied by a number of open source probabilistic programming packages that make them accessible to programmers and statisticians. PyMC3 is one such package written in Python and supported by NumFOCUS. This talk will give an introduction to probabilistic programming with PyMC3. No preexisting knowledge of Bayesian statistics is necessary; a working knowledge of Python will be helpful…more details

Talk | Research Frontiers

At Bloomberg, our research scientists and engineers work on a diverse set of data science problems — ranging from machine learning, natural language processing, dialog systems, information extraction, knowledge graphs, question answering, and table understanding. Due to the unique confluence of challenges we face at scale in terms of the data, precision, recall, and latency requirements of the global capital markets, we must build innovative solutions to rapidly deliver information to our clients across finance, business, government and philanthropy. These innovations take the shape of numerous publications in peer-reviewed conferences and journals, collaborations with academics and students, as well as partnerships with fellow researchers in other companies.

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 in applying AI, machine learning, natural language processing and deep learning across 5 key problem sets: making sense of news & social media, natural documents, signals from the financial markets, communication data and the impact of machine learning on finance...more details

Talk | Machine Learning | Beginner-Intermediate

Customer churn is the bane of online businesses, and data can help you understand the causes of churn and take action to reduce it. But there are many misconceptions about churn and pitfalls in actually using data driven insights to reduce churn. Usually, a standard predictive churn model provides little benefit! This is because interventions to reduce churn must be targeted by the causes of churn and those interventions are undertaken by diverse actors within the organization, all using separate tools and processes. The solution is for the data scientist to focus on feature engineering and to scientifically discover a powerful set of customer metrics that explain churn and engagement in a way that is interpretable to the business people. With great customer metrics, all teams focused on churn will speak the same language and be able to devise effective segments for interventions. Investment in feature engineering will also lead to an interpretable predictive model that is as accurate as possible, up to the limits imposed by the noisy and unbalanced nature of churn…more details

Talk | NLP | Open-source | Intermediate

Natural language processing has exploded in popularity during the last decade. No longer confined to academia, many companies now see NLP as a critical portion of their business intelligence, with the NLP market size expected to double again in the next two years. Traditional NLP approaches like sentiment analysis and topic modeling provide undeniably meaningful insights, but what other techniques can be leveraged to mine information from text?

This talk focuses on lesser known NLP methods that can help unearth novel observations and make analyses more memorable. After a brief introduction to the topic, attendees will learn about various open-source Python packages they can apply to enhance their NLP workflows. Example use cases will also be discussed to further solidify how each technique may be leveraged with existing data. Attendees of this talk will discover several unconventional NLP tools such as:
– Scattertext for comparing word usage between two populations
– spaCy’s linguistic features to parse sentences by syntax
– DeepMoji for assigning emoji labels to short text…more details

Talk | Machine Learning | ML for Programmers | Intermediate

The manufacturing industry is going through it’s IV industrial revolution where machines are connected, and data is harvested to discover deeper insights and solved problems to achieve a competitive edge in the market. There are various applications in industry 4.0, such as machine connectivity, machine productivity monitoring, predictive maintenance, predictive quality, inventory optimization, supply chain optimization, data discovery, and the possibilities are unlimited. Hence, there is a significant need for more reliable and robust approaches to achieve the benefits of various applications. By integrating the philosophy of lean manufacturing and “A small success eventually leads to bigger success,” one case of predictive maintenance (PdM) led to a global scale implementation by developing new technologies using ML & DL models and deploying them on an industrial scale. This case study discusses different levels of analytics, basics of predictive maintenance, anomaly detection, remaining useful life prediction techniques, combining data and SME knowledge, ML models, and how they can be deployed in manufacturing for different use cases. This case study also discusses various modeling and deployment challenges of implementing predictive maintenance in manufacturing…more details

Talk | MLOps & Data Engineering | Machine Learning | Intermediate

In this session, Oliver will walk through some of the best serverless options on how to operationalize ML pipelines within the Tensorflow ecosystem and on Google Cloud Platform, based on actual case studies. One of these real-life case studies will dive into the journey of a global cosmetics brand to become packaging-free with the help of ML. The first step towards this goal allows customers to view product information simply by taking a picture. This completely eliminates the need for packaging and labels in stores. However, in order to do this effectively, an accurate image classification model, accessible on mobile phones, is needed. This session will cover the details of the end-to-end machine learning pipeline that was created to deliver and update performance ML models to mobile users…more details

Talk | Computer Vision | Deep Learning | Intermediate-Advanced

ShopRunner is an e-commerce company that receives feeds of product data from many different retailer partners, including large department stores and retailers that specialize in apparel, electronics, nutritional products, and more. This data includes both images and text fields like product name and description. For apparel, which I will be focusing on in this talk, these attributes include color, pattern, sleeve length, season, fabric type, and country of origin. In typical retail settings such classifications are all performed manually, but at ShopRunner we have been developing deep learning models to automate the vast majority of this work.

In this talk I will motivate the benefits of combining image and text data for classification problems, and introduce our newly open sourced library for building multi-task computer vision and natural language processing deep learning pipelines using PyTorch. Using a variety of real world examples, I will walk through multiple strategies that we have employed at ShopRunner for combining images and text to classify apparel products. This will include an overview of the basics of transfer learning, a discussion of evaluation metrics and their tradeoffs in a business setting, and the overall design of our production product catalog enrichment system…more details

Talk | ML for Programmers | Kick-starter | Beginner

In this talk we’ll go over how to write code that is reproducible and easy for other people to work with.

We’ll start by talking about virtual environments. Virtual environments allow you to define the dependencies for your projects (such as NumPy or Matplotlib) and to keep these dependencies separated between projects. We’ll also outline some choices you have about how to manage your virtual environments.

Next we’ll talk about version control and why you should be using it even if you’re the only contributor to a project. Version control helps create a log of what work was done and why, and will give you the ability to go back when you inevitably make a change to your project that you can’t figure out how to undo.

Then we’ll discuss project structure by reviewing DrivenData’s Cookiecutter Data Science template. The template encourages a number of best practices, and makes it so that anyone familiar with the template will be able to look at your code for the first time will be reasonably well oriented.

Finally, we’ll briefly cover why you should establish coding styles and always use a linter...more details

Talk | MLOps & Data Engineering | ML for Programmers | Beginner-Intermediate

During this talk Tomek Urbaszek and Jarek Potiuk, both Apache Airflow official committers, will provide extensive introduction to Airflow. First part of the lecture will cover the most important components of Airflow. Attendees will become familiar with the definition of Directed Acyclic Graphs, task operators, integration hooks, tasks executors and the scheduler. This will help the audience to better understand the underlying concepts of Apache Airflow.

Second part of the talk will cover practical aspects of using Airflow. It will treat about best practices for developing workflows and implementing custom operators. This knowledge should help attendees to avoid potential pitfalls. Finally speakers will provide a guide to when Airflow is the right choice and when other solutions should be considered.

The talk is for all those who want to learn more about Apache Airflow. To fully appreciate the talk attendees should be familiar with basic concepts of scheduling work and distributed systems…more details

Track Keynote | Research Frontiers | Data for good | All Levels

Here we present and evaluate our machine learning approach to the Rodeo and release our SubseasonalRodeo dataset, collected to train and evaluate our forecasting system.

Our system is an ensemble of two nonlinear regression models. The first integrates the diverse collection of meteorological measurements and dynamic model forecasts in the SubseasonalRodeo dataset and prunes irrelevant predictors using a customized multitask model selection procedure. The second uses only historical measurements of the target variable (temperature or precipitation) and introduces multitask nearest neighbor features into a weighted local linear regression. Each model alone is significantly more accurate than the debiased operational U.S. Climate Forecasting System (CFSv2), and our ensemble skill exceeds that of the top Rodeo competitor for each target variable and forecast horizon. Moreover, over 2011-2018, an ensemble of our regression models and debiased CFSv2 improves debiased CFSv2 skill by 40-50% for temperature and 129-169% for precipitation. We hope that both our dataset and our methods will help to advance the state of the art in subseasonal forecasting…more details

Talk | Deep Learning | Intermediate-Advanced

What if you could process complex deep learning models at scale without specialized AI hardware? Award-winning MIT mathematics professor Nir Shavit will demonstrate how to make inference scale in the real world using only software on commodity CPUs. The Neural Magic Inference Engine bucks conventional wisdom about high-throughput computing, and may change your mind about machine learning forever. For nearly a decade, data scientists have believed that deep learning models need the parallel computing power of a GPU to get “good enough” results. GPUs, however, have limited cache memory, which requires a data scientist to sacrifice accuracy for performance gains to achieve inference at scale. CPUs, by contrast, have much more sophisticated cache memory hierarchies that are perfect for addressing these constraints. This session will show how to run deep neural networks at scale with GPU-class performance on commodity CPUs, with all of the deployment flexibility of a software solution. Data science teams can run models at exceptional speeds, without the expense and complexity of dedicated hardware…more details