Sunil Kumar Vuppala of Philips Research at Machine Learning Developers Summit 2019
summary
Describes methods to speed up inference of deep learning models used in production, focusing on practical techniques such as model compression, quantization, pruning, and optimized runtime implementations. Covers trade-offs between latency, throughput and accuracy, and discusses hardware-aware optimizations for CPUs, GPUs and edge devices.
AI-written summary · about
related talks
24mTools & Techniques for a cost optimized Analytics & AI pipeline
32mTaming the Machine: Basics of ML Models Training and Inference Optimization
37mDeep Learning Neural Network Acceleration at the Edge
15mDesign Compact Deep Learning Models: Small is the New Big
14mOptimize your TensorFlow Lite models
29mRunning Deep Learning Models on Mobile with optimised speed
16mLightning Talk: Efficient Inference at the Edge: Performance You Need at the Lowest...
21mThings I learned running neural networks on microcontrollers