Skip to content
lectern.tv
Stanford CS231n: Convolutional Neural Networks for Visual Recognition →0/16
1

Lecture 1 | Introduction to Convolutional Neural Networks for Visual Recognition

57:56
2

Lecture 2 | Image Classification

59:31
3

Lecture 3 | Loss Functions and Optimization

1:14:40
4

Lecture 4 | Introduction to Neural Networks

1:13:59
5

Lecture 5 | Convolutional Neural Networks

1:08:56
6

Lecture 6 | Training Neural Networks I

1:20:19
7

Lecture 7 | Training Neural Networks II

1:15:30
8

Lecture 8 | Deep Learning Software

1:18:07
9

Lecture 9: CNN Architectures (CS231n)

1:17:40
10

Lecture 10: Recurrent Neural Networks (CS231n)

1:13:09
11

Lecture 11: Detection and Segmentation (CS231n)

1:14:26
12

Lecture 12: Visualizing and Understanding (CS231n)

1:15:48
13

Lecture 13: Generative Models (CS231n)

1:17:41
14

Lecture 14: Deep Reinforcement Learning (CS231n)

1:04:01
15

Lecture 15: Efficient Methods and Hardware for Deep Learning (CS231n)

1:16:52
16

Lecture 16: Adversarial Examples and Adversarial Training (CS231n)

1:21:46

Lecture 15: Efficient Methods and Hardware for Deep Learning (CS231n)

Song HanStanford University

Professor

1:16:52·2017·graduate level

Guest lecturer Song Han presents algorithms for accelerating deep learning inference and training, including pruning, quantization, and parallelization. The lecture also covers specialized hardware like GPUs, FPGAs, ASICs, and TPUs.

Systems & CSAI & MLDeep LearningComputer ArchitectureAlgorithms & Data StructuresModel Pruning

In Lecture 15, guest lecturer Song Han discusses algorithms and specialized hardware that can be used to accelerate training and inference of deep learning workloads. We discuss pruning, weight sharing, quantization, and other techniques for accelerating inference, as well as parallelization, mixed precision, and other techniques for accelerating training. We discuss specialized hardware for deep learning such as GPUs, FPGAs, and ASICs, including the Tensor Cores in NVIDIA’s latest Volta GPUs as well as Google’s Tensor Processing Units (TPUs).

Keywords: Hardware, CPU, GPU, ASIC, FPGA, pruning, weight sharing, quantization, low-rank approximations, binary networks, ternary networks, Winograd transformations, EIE, data parallelism, model parallelism, mixed precision, FP16, FP32, model distillation, Dense-Sparse-Dense training, NVIDIA Volta, Tensor Core, Google TPU, Google Cloud TPU

Previous · Lecture 14

Lecture 14: Deep Reinforcement Learning (CS231n)

Next · Lecture 16

Lecture 16: Adversarial Examples and Adversarial Training (CS231n)

Seminar notes & discussion

Sign in to comment

No public seminar notes on this lecture yet.