xianglinyang

DeepVisualInsight

DeepVisualInsight (DVI) is a time-travelling visualization solution for how the classification boundaries of a learned model is trained in two-dimensional space.

TimeVis

Official source code for IJCAI 2022 Paper: Temporality Spatialization: A Scalable and Faithful Time-Travelling Visualization for Deep Classifier Training

DeepDebugger

Official source code for ESEC/FSE 2023 Paper: DeepDebugger: An Interactive Time-Travelling Debugging Approach for Deep Classifiers

ActiveLearning

active learning SOTA

AnalogicalReasoners

TS

Simple proof minimax optimal thompson sampling.

SentryCam

A real-time visualization for monitoring the training of DNNs.

backdoor_attack

Backdoor Attacks

backdoor_attack_nlp

An open-source toolkit for textual backdoor attack and defense (NeurIPS 2022 D&B, Spotlight)

continual-learning

PyTorch implementation of various methods for continual learning (XdG, EWC, SI, LwF, FROMP, DGR, BI-R, ER, A-GEM, iCaRL, Generative Classifier) in three different scenarios.

DeepLearning

深度学习入门教程, 优秀文章, Deep Learning Tutorial

DeepViewExp

DeepVis

DVI_baseline_exp

DVI_supplementary

the experiment code for DVI

EMNIST_Resnet

subject models for DVI using EMNIST dataset

example_cml

github-actions-tutorial

Sample repository to demonstrate use of Github Actions for SE&PL Research Group.

gitignore

A collection of useful .gitignore templates

intrinsic-dimensionality

PyTorch Implementation of CVPR'19 - On the Intrinsic Dimensionality of Image Representation

Noisy_Training

Noisy datset. The case study for dl-vis-debugger.

PyG-GNN

graph dataset for visualization

pytorch-cifar

95.47% on CIFAR10 with PyTorch

PyTorch_CIFAR10

Pretrained TorchVision models on CIFAR10 dataset (with weights)

tensorboard

TensorFlow's Visualization Toolkit

tensorboardVis

vis_datasets

The subject model datasets for visualization methods.

xianglinyang

xianglinyang.github.io

xianglin's homepage

YOPO-You-Only-Propagate-Once

Code for our nips19 paper: You Only Propagate Once: Accelerating Adversarial Training Via Maximal Principle