Google's Machine Learning Virtual Community Day 2021
2021-11-09
-- online --
as the organisers put it
Google の機械学習プロダクトに関する最新の発表にご参加ください。コミュニティからのハイライトと、Google の人工知能と機械学習テクノロジーのエコシステムに参加するための方法を共有します。
who it's for
machine learning engineers, research scientists, ml infrastructure engineers, mobile machine learning engineers, mlops engineers, ai ethics practitioners, developer advocates
topics
JAX automatic differentiation and jit; TensorFlow Lite on-device training; MoveNet multi-person pose detection; Cloud TPU v4 architecture and pods; ParameterServerStrategy distributed training; TFX production pipelines and components; Responsible AI Toolkit with LIT and TCAV; AlphaFold protein structure prediction; chip floorplanning with deep reinforcement learning; TensorFlow.js browser model training; Vertex AI managed training and serving; model quantization and XNNPack optimizations
| title | author | length |
|---|---|---|
| Machine learning tools from Google | Nadav Eiron | 07m |
TensorFlow updates focus on concrete tools and numbers: TF 2.4 adds MultiWorkerMirroredStrategy and mixed precision for faster GPU and TPU training, TF 2.6… | ||
| Machine Learning for mobile | Kemal El Moujahid | 10m |
ML Kit offers 13 production-ready on-device APIs, and TF Lite runs on more than 4 billion devices while adding on-device training and a Play Services runtime… | ||
| State-of-the-art Machine Learning research with Google tools | Laurence Moroney | 04m |
Cloud-managed TPUs scale into TPU Pods that run state-of-the-art training and inference, while T5X provides a modular framework for configurable sequence-model… | ||
| How Machine Learning tools at Google made AlphaFold possible | Tim Green | 04m |
AlphaFold uses deep learning with attention mechanisms to predict protein 3D structure from sequence and is judged at CASP as a general solution to the folding… | ||
| Engage with the Machine Learning community | Jeanine Banks | 06m |
Overview of concrete community paths: join 175 Google Developer Experts offering talks and workshops, or one of 87 TensorFlow user groups in 37 countries; the… | ||
| Simplified Machine Learning on mobile, browsers, and edge devices with on-device ML | Yu-Cheng Ling, Arun Venkatesan | 15m |
Demonstrates true on-device training with a shipped Android demo that collects images and trains a classifier entirely on phone using TensorFlow Lite and… | ||
| Manage MLOps and deploy Machine Learning to production with the new and improved TFX | Robert Crowe, Tris Warkentin | 08m |
TFX is a production MLOps stack that claims model code is only 5% of total code and focuses on pipelines, portability (Colab, notebook, Kubernetes, Vertex)… | ||
| Building fair, ethical, and responsible AI with the Responsible AI Toolkit | Parker Barnes, Ludovic Peran | 12m |
Google’s Responsible AI Toolkit provides concrete tools and results: use Fairness Indicators to slice a restaurant-review sentiment model and find lower… | ||
| Intro to JAX: Accelerating Machine Learning research | Jake VanderPlas | 11m |
JAX traces Python numerical functions into a jaxpr intermediate representation, applies composable transforms like grad, jit and vmap to automatically… | ||
| Simplified distributed training with tf.distribute parameter servers | Yuefeng Zhou | 05m |
Explains how ParameterServerStrategy gives asynchronous, data-parallel training for large-embedding or preemptible-machine workloads, using a single-client… | ||
| Cloud TPU v4: Fast, flexible, and easy-to-use ML accelerators | Shauheen Zahirazami | 06m |
Cloud TPU v4 doubles per-chip speed versus v3, packs 4,096 v4 chips into a pod with 10x interconnect bandwidth per chip and delivers more than one exaflop of… | ||
| Chip floorplanning with deep reinforcement learning | Anna Goldie, Azalia Mirhoseini | 10m |
A deep reinforcement learning pipeline generates chip floorplans for Google TPUs that match or beat expert layouts and runs under six hours instead of weeks… | ||
| Incredible things your community is creating with TensorFlow.js | Jason Mayes | 05m |
YogaAI classifies six poses and runs a 10-second hold timer that pauses when detection fails; PoseNet extracts body keypoints and extrudes pose data into 3D/AR… | ||
| Building Machine Learning models with Edge Impulse | Josh Gordon, Jenny Plunkett, Jan Jongboom | 05m |
Edge Impulse is presented as an end-to-end TinyML platform with 40,000 projects in the community, a web IDE that ingests sensor or audio data via QR-collected… | ||
| How to get involved in Machine Learning | Joana Carrasqueira | 08m |
Creates a TensorFlow account and joins the TensorFlow Forum (2,500+ members) to follow getting_started and community tags, or contributes code via the RFCs… | ||
| Powered by TensorFlow: How machine learning helps advance access to human rights information | NICOLETTA ZAPILLE | 03m |
HURIDOCS and UPR Info use BERT on TensorFlow Hub to build a classifier that assigns human-rights topics to UPR recommendations, trained on 60,000 hand-labeled… | ||
| Powered by TensorFlow: Creating a custom, machine learning-powered drumming arm | Jason Barnes | 03m |
A custom robotic drumming prosthetic reads muscle signals, trains a deep model on recorded EMG data, and runs inference on a small console so the arm predicts… | ||
| nothing matches | ||