Angel 3.0 - A Full Stack Machine Learning Platform
Offered By: Linux Foundation via YouTube
Course Description
Overview
Explore a comprehensive overview of Angel 3.0, a full-stack machine learning platform, in this 41-minute conference talk by Fitz Wang from Tencent. Delve into the platform's architecture, key features, and real-world applications. Learn about auto feature engineering, Spark On Angel integration, hyperparameter tuning, and model serving capabilities. Discover how Angel 3.0 completes the machine learning pipeline with components like feature engineering, model training, and hyperparameter optimization. Gain insights into the platform's support for huge recommendation models, sparse input data, and cross-platform model serving. Understand the integration of PyTorch for gradient computation and Angel's parameter server for efficient parameter management. Explore graph embedding and GNN algorithms, as well as Spark 2.4 and Kubernetes adaptations. Examine practical use cases, including short video recommendations and financial anti-fraud applications at Tencent.
Syllabus
Intro
What is Angel? Angel is a large scale distributed machine learning platform
Angel Architecture
Angel outside Tencent
Open source & Papers
Angel 3.0 features
Auto Feature Engineering (1/2)
Spark On Angel (SONA, 2/3)
Hyper Parameter Tuning(1/2)
Angel Serving
Short Video Recommendation in Tencent
Financial Anti-Fraud in Tencent
Taught by
Linux Foundation
Tags
Related Courses
Bite-Sized Neo4j for Data ScientistsYouTube Graph Convolutional Networks - GNN Paper Explained
Aleksa Gordić - The AI Epiphany via YouTube Discriminative Prototype Selection for Graph Embedding
University of Central Florida via YouTube Building a Knowledge Graph with Spark and NLP for Novel Drug Recommendations
Databricks via YouTube Inapproximability of Clustering in Lp Metrics
IEEE via YouTube