O'Reilly TensorFlow World

"Human error": How can we help people build models that do what they expect

February 24, 2020

It's never been easier to train machine learning models. With excellent open source tooling, lower compute techniques, and incredible educational material online, really anybody can start to train their own models today. Yet, Anna Roth explains, when domain experts try to transfer their expertise to an ML model, the results can be unpredictable.

A novel solution for a data augmentation and bias problem in NLP using TensorFlow

February 24, 2020

Join KC Tung to discover a way to use TensorFlow to solve a natural language processing (NLP) model bias problem with data augmentation for an enterprise customer (one of the largest airlines in the world). KC leveraged hidden gems in tf.data and the new API to easily find a novel use for text generation and found it surprisingly improved his NLP model.

Accelerating ML at Twitter

February 24, 2020

Twitter employs ML throughout its product to deliver value for its customers. Theodore Summe gives you a glimpse into ML at Twitter and explains how Cortex works to accelerate ML to better serve customer needs by partnering with TensorFlow.

Accelerating training, inference, and ML applications on NVIDIA GPUs

February 24, 2020

Maggie Zhang, Nathan Luehr, Josh Romero, Pooya Davoodi, and Davide Onofrio give you a sneak peek at software components from NVIDIAs software stack so you can get the best out of your end-to-end AI applications on modern NVIDIA GPUs. They also examine features and tips and tricks to optimize your workloads right from data loading, processing, training, inference, and deployment.

Advanced model deployments with TensorFlow Serving

February 24, 2020

Hannes Hapke leads a deep dive into deploying TensorFlow models within minutes with TensorFlow Serving and optimizing your serving infrastructure for optimal throughput.

Deep Learning Toolkit for Splunk

February 24, 2020

The Deep Learning Toolkit for Splunk allows you to integrate advanced custom machine learning systems with the Splunk platform using TensorFlow 2.0, PyTorch and NLP libraries. Jupyter Lab Notebooks are providing data scientists and machine learning developers with an integrated experience from rapid prototyping to operationalising models in production. The app is freely available on splunkbase.

Dont beat the market; beat the bots: Adversarial networks in finance

February 24, 2020

Automated investing has brought an immense amount of stability to the market, but it has also brought predictability. Garrett Lander and Al Kari examine if an adversarial network can game the behavior of automated investors by learning the patterns in market activity to which they are most vulnerable.

EasyAgents: Reinforcement Learning for people who want to solve real-world problems

February 24, 2020

Reinforce Learning can be a game changer when you do not have training data, but are instead able to simulate an environment. Unfortunately, the theory of Reinforcement Learning is complex and the vast number of algorithms in that area adds to the burden for getting started. Easyagents takes some of the burden by making it a one-liner to run a Reinforcement Learning algorithm on your problem.

Effective sampling methods within TensorFlow input functions

February 24, 2020

Many real-world machine learning applications require generative or reductive sampling of data. Laxmi Prajapat and William Fletcher demonstrate sampling techniques applied to training and testing data directly inside the input function using the tf.data API.

Generative malware outbreak detection

February 23, 2020

Practical defense systems require precise detection during malware outbreaks with only a handful of available samples. Sean Park demonstrates how to detect in-the-wild malware samples with a single training sample of a kind, with the help of TensorFlow's flexible architecture in implementing a novel variable-length generative adversarial autoencoder.

HARP: An efficient and elastic GPU-sharing system

February 23, 2020

Pengfei Fan and Lingling Jin offer an overview of an efficient and elastic GPU-sharing system for users who do research and development with TensorFlow.

How Criteo optimized and sped up its TensorFlow models by 10x and served them under 5 ms

February 23, 2020

Criteo's real-time bidding of ad spaces requires its TensorFlow (TF) models to make online predictions in less than 5 ms. Nicolas Kowalski and Axel Antoniotti explain why Criteo moved away from high-level APIs and rewrote its models from scratch, reimplementing cross-features and hashing functions using low-level TF operations in order to factorize as much as possible all TF nodes in its model.

Introduction to Hilbert AutoML with TensorFlow Extended (TFX) at Yahoo! JAPAN

February 23, 2020

Hilbert is an AI framework that works with TensorFlow Extended (TFX) at Yahoo! JAPAN, which provides AutoML to create production-level deep learning models automatically. Hilbert is currently used by over 20 services of Yahoo! JAPAN. Shin-Ichiro Okamoto details how to achieve production-level AutoML and explores service use cases at Yahoo! JAPAN.

Machine learning over real-time streaming data with TensorFlow

February 23, 2020

In many applications where data is generated continuously, combining machine learning with streaming data is imperative to discover useful information in real time. Yong Tang explores TensorFlow I/O, which can be used to easily build a data pipeline with TensorFlow and stream frameworks such as Apache Kafka, AWS Kinesis, or Google Cloud PubSub.

MLIR: Accelerating AI

February 23, 2020

MLIR is TensorFlow's open source machine learning compiler infrastructure that addresses the complexity caused by growing software and hardware fragmentation and makes it easier to build AI applications. Chris Lattner and Tatiana Shpeisman explain how MLIR is solving this growing hardware and software divide and how it impacts you in the future.

Modular convolution considered beneficial

February 23, 2020

Jack Chung, Chao Liu, and Daniel Lowell explore breaking convolution algorithms into modular pieces to be better fused with graph compilers such as accelerated linear algebra (XLA).

Natural language processing using transformer architectures

February 23, 2020

Transformer architectures have taken the field of natural language processing (NLP) by storm and pushed recurrent neural networks to the sidelines. Aurlien Gron examines transformers and the amazing language models based on them (e.g., BERT and GPT 2) and shows how you can use them in your projects.

Node-RED and TensorFlow.js: Developing deep learning IoT apps in the browser

February 23, 2020

Va Barbosa and Paul Van Ec highlight the benefits of using TensorFlow.js and Node-RED together as an educational tool to engage developers and provide you with a powerful, creativity-inspiring platform for interacting and developing with machine learning models.

Opening keynote

February 23, 2020

Jeff Dean explains why Google originally open-sourced TensorFlow almost four years ago. Join in to learn about TensorFlow's progress and how it can solve the problems you care about.

Production ML pipelines with TensorFlow Extended (TFX)

February 23, 2020

Aurlien Gron dives into creating production ML pipelines with TensorFlow Extended (TFX) and using TFX to move from ML coding to ML engineering. You'll walk through the basics and put your first pipeline together, then learn how to customize TFX components and perform deep analysis of model performance.

TensorFlow community announcements

February 23, 2020

Kemal El Moujahid divulges exciting developments for the TensorFlow community. Join in to learn how the TensorFlow team provides new and improved resources for developers and enterprises to succeed.

TensorFlow Hub: The platform to share and discover pretrained models for TensorFlow

February 23, 2020

Machine learning is a difficult skill to master for the many developers who are starting to use TensorFlow. Many developers use TensorFlow today, yet the majority of software developers out there have yet to learn machine learning. Mike Liang takes you through TensorFlow Hub, designed to help developers make better and faster user of machine learning in their products.

TensorFlow, open source, and IBM (sponsored by IBM)

February 22, 2020

IBM has a long history of contributing to the open source projects that make the most difference to its clients, and the company has been working to build responsible solutions to enterprise data science problems for many years. Join Frederick Reiss to hear about IBM's role in open source software, TensorFlow, building AI solutions, and what IBM is excited about with this latest (2.0) release.

TensorFlow.js: Bringing machine learning to JavaScript

February 22, 2020

JavaScript is the most widely used programming language in the world, and with TensorFlow.js, you can bring the power of TensorFlow and machine learning to your JavaScript application. Sandeep Gupta and Joseph Paul Cohen introduce the TensorFlow.js library and showcase the amazing possibilities of combining machine learning with JavaScript-based web, mobile, and server-side applications.

Zero to ML hero with TensorFlow 2.0

February 22, 2020

Get a programmer's perspective on machine learning with Laurence Moroney, from the basics all the way up to building complex computer vision scenarios using convolutional neural networks and natural language processing with recurrent neural networks.