Official implementation of Representer Point Selection via Local Jacobian Expansion for Post-hoc Classifier Explanation of Deep Neural Networks and Ensemble Models at NeurIPS 2021

Related tags

Deep LearningRPS_LJE
Overview

Representer Point Selection via Local Jacobian Expansion for Classifier Explanation of Deep Neural Networks and Ensemble Models

This repository is the official implementation of Representer Point Selection via Local Jacobian Expansion for Classifier Explanation of Deep Neural Networks and Ensemble Models at NeurIPS 2021. (will update the link)

Introduction

We propose a novel sample-based explanation method for classifiers with a novel derivation of representer point with Taylor Expansion on the Jacobian matrix.

If you would like to cite this work, a sample bibtex citation is as following:

@inproceedings{yi2021representer,
 author = {Yi Sui, Ga Wu, Scott Sanner},
 booktitle = {Advances in Neural Information Processing Systems},
 title = {Representer Point Selection via Local Jacobian Expansion for Classifier Explanation of Deep Neural Networks and Ensemble Models},
 year = {2021}
}

Set up

To install requirements:

pip install -r requirements.txt

Change the root path in config.py to the path to the project

project_root = #your path here

Download the pre-trained models and calculated weights here

  • Dowload and unzip the saved_models_MODEL_NAME
  • Put the content into the corresponding folders ("models/ MODEL_NAME /saved_models")

Training

In our paper, we run experiment with three tasks

  • CIFAR image classification with ResNet-20 (CNN)
  • IMDB sentiment classification with Bi-LSTM (RNN)
  • German credit analysis with XGBoost (Xgboost)

The models are implemented in the models directory with pre-trained weights under "models/ MODEL_NAME /saved_models/base" : ResNet (CNN), Bi-LSTM (RNN), and XGBoost.

To train theses model(s) in the paper, run the following commands:

python models/CNN/train.py --lr 0.01 --epochs 10 --saved_path saved_models/base
python models/RNN/train.py --lr 1e-3 --epochs 10 --saved_path saved_models/base --use_pretrained True
python models/Xgboost/train.py

Caculate weights

We implemented three different explainers: RPS-LJE, RPS-l2 (modified from official repository of RPS-l2), and Influence Function. To calculate the importance weights, run the following commands:

python explainer/calculate_ours_weights.py --model CNN --lr 0.01
python explainer/calculate_representer_weights.py --model RNN --lmbd 0.003 --epoch 3000
python explainer/calculate_influence.py --model Xgboost

Experiments

Dataset debugging experiment

To run the dataset debugging experiments, run the following commands:

python dataset_debugging/experiment_dataset_debugging_cnn.py --num_of_run 10 --flip_portion 0.2 --path ../models/CNN/saved_models/experiment_dataset_debugging --lr 1e-5
python dataset_debugging/experiment_dataset_debugging_cnn.py --num_of_run 10 --flip_portion 0.2 --path ../models/CNN/saved_models/experiment_dataset_debugging_fix_random_split --lr 1e-5 --seed 11

python dataset_debugging/experiment_dataset_debugging_rnn.py --num_of_run 10 --flip_portion 0.2 --path ../models/RNN/saved_models/experiment_dataset_debugging --lr 1e-5

python dataset_debugging/experiment_dataset_debugging_Xgboost.py --num_of_run 10 --flip_portion 0.3 --path ../models/Xgboost/saved_models/experiment_dataset_debugging --lr 1e-5

The trained models, intermediate outputs, explainer weights, and accuracies at each checkpoint are stored under the specified paths "models/MODEL_NAME/saved_models/experiment_dataset_debugging". To visualize the results, run the notebooks plot_res_cnn.ipynb, plot_res_cnn_fixed_random_split.ipynb, plot_res_rnn.ipynb, plot_res_xgboost.ipynb. The results are saved under folder dataset_debugging/figs.

Other experiments

All remaining experiments are in Jupyter-notebooks organized under "models/ MODEL_NAME /experiments" : ResNet (CNN), Bi-LSTM (RNN), and XGBoost.

A comparison of explanation provided by Influence Function, RPS-l2, and RPS-LJE. Explanation for Image Classification

Owner
Yi(Amy) Sui
Yi(Amy) Sui
PyTorch implementation of "ContextNet: Improving Convolutional Neural Networks for Automatic Speech Recognition with Global Context" (INTERSPEECH 2020)

ContextNet ContextNet has CNN-RNN-transducer architecture and features a fully convolutional encoder that incorporates global context information into

Sangchun Ha 24 Nov 24, 2022
Aerial Imagery dataset for fire detection: classification and segmentation (Unmanned Aerial Vehicle (UAV))

Aerial Imagery dataset for fire detection: classification and segmentation using Unmanned Aerial Vehicle (UAV) Title FLAME (Fire Luminosity Airborne-b

79 Jan 06, 2023
Official Pytorch Implementation of Adversarial Instance Augmentation for Building Change Detection in Remote Sensing Images.

IAug_CDNet Official Implementation of Adversarial Instance Augmentation for Building Change Detection in Remote Sensing Images. Overview We propose a

53 Dec 02, 2022
A video scene detection algorithm is designed to detect a variety of different scenes within a video

Scene-Change-Detection - A video scene detection algorithm is designed to detect a variety of different scenes within a video. There is a very simple definition for a scene: It is a series of logical

1 Jan 04, 2022
Ranger - a synergistic optimizer using RAdam (Rectified Adam), Gradient Centralization and LookAhead in one codebase

Ranger-Deep-Learning-Optimizer Ranger - a synergistic optimizer combining RAdam (Rectified Adam) and LookAhead, and now GC (gradient centralization) i

Less Wright 1.1k Dec 21, 2022
This code provides various models combining dilated convolutions with residual networks

Overview This code provides various models combining dilated convolutions with residual networks. Our models can achieve better performance with less

Fisher Yu 1.1k Dec 30, 2022
Codes for "Solving Long-tailed Recognition with Deep Realistic Taxonomic Classifier"

Deep-RTC [project page] This repository contains the source code accompanying our ECCV 2020 paper. Solving Long-tailed Recognition with Deep Realistic

Gina Wu 16 May 26, 2022
Dense Contrastive Learning (DenseCL) for self-supervised representation learning, CVPR 2021.

Dense Contrastive Learning for Self-Supervised Visual Pre-Training This project hosts the code for implementing the DenseCL algorithm for se

Xinlong Wang 491 Jan 03, 2023
Bianace Prediction Pytorch Model

Bianace Prediction Pytorch Model Main Results ETHUSDT from 2021-01-01 00:00:00 t

RoyYang 4 Jul 20, 2022
本项目是一个带有前端界面的垃圾分类项目,加载了训练好的模型参数,模型为efficientnetb4,暂时为40分类问题。

说明 本项目是一个带有前端界面的垃圾分类项目,加载了训练好的模型参数,模型为efficientnetb4,暂时为40分类问题。 python依赖 tf2.3 、cv2、numpy、pyqt5 pyqt5安装 pip install PyQt5 pip install PyQt5-tools 使用 程

4 May 04, 2022
Processed, version controlled history of Minecraft's generated data and assets

mcmeta Processed, version controlled history of Minecraft's generated data and assets Repository structure Each of the following branches has a commit

Misode 75 Dec 28, 2022
SAPIEN Manipulation Skill Benchmark

ManiSkill Benchmark SAPIEN Manipulation Skill Benchmark (abbreviated as ManiSkill, pronounced as "Many Skill") is a large-scale learning-from-demonstr

Hao Su's Lab, UCSD 107 Jan 08, 2023
Diverse Branch Block: Building a Convolution as an Inception-like Unit

Diverse Branch Block: Building a Convolution as an Inception-like Unit (PyTorch) (CVPR-2021) DBB is a powerful ConvNet building block to replace regul

253 Dec 24, 2022
A transformer model to predict pathogenic mutations

MutFormer MutFormer is an application of the BERT (Bidirectional Encoder Representations from Transformers) NLP (Natural Language Processing) model wi

Wang Genomics Lab 2 Nov 29, 2022
An implementation for the loss function proposed in Decoupled Contrastive Loss paper.

Decoupled-Contrastive-Learning This repository is an implementation for the loss function proposed in Decoupled Contrastive Loss paper. Requirements P

Ramin Nakhli 71 Dec 04, 2022
Supercharging Imbalanced Data Learning WithCausal Representation Transfer

ECRT: Energy-based Causal Representation Transfer Code for Supercharging Imbalanced Data Learning With Energy-basedContrastive Representation Transfer

Zidi Xiu 11 May 02, 2022
StableSims is an open-source project aimed at simulating MakerDAO's Dai stablecoin system

StableSims is an open-source project aimed at simulating MakerDAO's Dai stablecoin system, initially used for researching optimal incentive parameters for Liquidations 2.0.

Blockchain at Berkeley 52 Nov 21, 2022
Install alphafold on the local machine, get out of docker.

AlphaFold This package provides an implementation of the inference pipeline of AlphaFold v2.0. This is a completely new model that was entered in CASP

Kui Xu 73 Dec 13, 2022
ThunderSVM: A Fast SVM Library on GPUs and CPUs

What's new We have recently released ThunderGBM, a fast GBDT and Random Forest library on GPUs. add scikit-learn interface, see here Overview The miss

Xtra Computing Group 1.4k Dec 22, 2022
pytorch implementation of the ICCV'21 paper "MVTN: Multi-View Transformation Network for 3D Shape Recognition"

MVTN: Multi-View Transformation Network for 3D Shape Recognition (ICCV 2021) By Abdullah Hamdi, Silvio Giancola, Bernard Ghanem Paper | Video | Tutori

Abdullah Hamdi 64 Jan 03, 2023