Track:

Deep Learning Applications & Practices

Location: Cyril Magnin I

Day of week: Wednesday

Deep learning lessons using Tensorflow, Keras, PyTorch, Caffe across machine translation, computer vision.

Track Host:
Soups Ranjan
Director of Data Science @Coinbase

Soups Ranjan is the Director of Data Science at Coinbase, one the largest bitcoin exchanges in the world. He manages the Risk & Data Science team that is chartered with preventing avoidable losses to the company due to payment fraud or account takeovers. Soups has a PhD in ECE on network security from Rice University. He has previously led the development of Machine Learning pipelines to improve performance advertising at Yelp and Flurry. He is the founder of RiskSalon.org, a round-table forum for risk professionals in San Francisco to share ideas on stopping bad actors.

SHORT TALK (10 MIN)

9:00am - 9:10am

PyTorch by Example

Jendrik Joerdening, Data Scientist @Aurubis
CASE STUDY TALK (50 MIN)

9:20am - 10:10am

Recent Deep Learning Advancements, Revolution Far from Over!

It started with Backpropagation, convolutional neural networks and RMSprop but these are only the first wave of innovations that got deep learning started. Some recent advancements include Population Based Training (PBT), Neuroevolution and ML Managed Services. We will discuss how deep learning research is becoming increasingly automated, more computationally efficient and serverless.

You will learn tips on how to automatically search the best deepnet topology for your task, scale up your machine learning experiments to run on thousands of nodes and to do hyperparameter optimization like pro!

Arshak Navruzyan, Founder @fellowshipai
SHORT TALK (10 MIN)

10:35am - 10:45am

Tensorflow Jumpstart

Magnus Hyttsten, TensorFlow Developer Advocate @Google
CASE STUDY TALK (50 MIN)

10:55am - 11:45am

TensorFlow: Pushing the ML Boundaries

Google has made extraordinary advances in Machine Learning (ML) over the past few years. These breakthroughs requires enormous amounts of computation, both to train as well as run the underlying machine learning models. That's why we've built and deployed the Tensor Processing Unit (TPU), to allow us to support larger and larger amounts of ML workloads. In this talk, we'll look at the architecture of the TPU, how you write effective TensorFlow code to utilize its power, and how Google is using it to break new frontiers in Machine Learning.

Magnus Hyttsten, TensorFlow Developer Advocate @Google
SHORT TALK (10 MIN)

12:45pm - 12:55pm

Basics of Deep Learning: No Math Required

Roland Meertens, Machine Learning Engineer @Autonomous Intelligent Driving
CASE STUDY TALK (50 MIN)

1:05pm - 1:55pm

Interpretable Machine Learning Products

Interpretable models are easier to improve. Regulators and society can better trust them to be safe and nondiscriminatory. They can also offer insights that can be used to change real-world outcomes for the better. But because there is a central tension between accuracy and interpretability interpretability can be hard to ensure.

I'll explore both the product case for interpretability and the academic research that is starting to make the inner workings of black box models such as deep neural networks easier to understand. In particular, I'll look at the application of a new open source tool called LIME to customer churn, image classification and black box NLP models.

Mike Lee Williams, Research engineer @Cloudera Fast Forward Labs
SHORT TALK (10 MIN)

2:20pm - 2:30pm

Detecting Similar Id Documents Using Deep Learning

Burkay Gur, Risk Engineer @Coinbase
CASE STUDY TALK (50 MIN)

2:40pm - 3:30pm

Understanding ML/DL Models using Interactive Visualization Techniques

In this talk we will see how visualization techniques can help us better understand machine learning and deep learning models. We will look at a specific financial use-case and see how interactive visualizations can help us not only improve the the model but also fix data issues. We will then take a look at tools to build, train and diagnose deep learning models applied on financial data sets.

Chakri Cherukuri, Quantitative Researcher @Bloomberg

Tracks

  • Deep Learning Applications & Practices

    Deep learning lessons using tooling such as Tensorflow & PyTorch, across domains like large-scale cloud-native apps and fintech, and tacking concerns around interpretability of ML models.

  • Predictive Data Pipelines & Architectures

    Best practices for building real-world data pipelines doing interesting things like predictions, recommender systems, fraud prevention, ranking systems, and more.

  • ML in Action

    Applied track demonstrating how to train, score, and handle common machine learning use cases, including heavy concentration in the space of security and fraud

  • Real-world Data Engineering

    Showcasing DataEng tech and highlighting the strengths of each in real-world applications.