implementation of the paper "MarginGAN: Adversarial Training in Semi-Supervised Learning"

Last update: Dec 23, 2022

Related tags

Overview

MarginGAN

This repository is the implementation of the paper "MarginGAN: Adversarial Training in Semi-Supervised Learning".

1."preliminary" is the implementation of Preliminary Experiment on MNIST of the paper. Thank the authors of pytorch-generative-model-collections and examples of pytorch, our code is widely adapted from their repositories.

To train the network, an example is as follows:

python main.py \
  --gan_type MarginGAN \
  --num_labels 600 \
  --lrC 0.1 \
  --epoch 50

2."ablation" is the implementation of Ablation Study on MNIST of the paper. To train the network, an example is as follows:

python main.py \
--gan_type MarginGAN_UG \
--num_labels 600 \
--lrC 0.01 \
--epoch 50

3."further" is the implementation of Experiment on SVHN and CIFAR-10 of the paper. Thank the authors of mean teacher, our code is widely adapted from their repositories.

To train the network, an example is as follows:

python MarginGAN_main.py \
    --dataset cifar10 \
    --train-subdir train+val \
    --eval-subdir test \
    --batch-size 128 \
    --labeled-batch-size 31 \
    --arch cifar_shakeshake26 \
    --consistency-type mse \
    --consistency-rampup 5 \
    --consistency 100.0 \
    --logit-distance-cost 0.01 \
    --weight-decay 2e-4 \
    --lr-rampup 0 \
    --lr 0.05 \
    --nesterov True \
    --labels data-local/labels/cifar10/1000_balanced_labels/00.txt  \
    --epochs 180 \
    --lr-rampdown-epochs 210 \
    --ema-decay 0.97 \
    --generated-batch-size 32

implementation of the paper "MarginGAN: Adversarial Training in Semi-Supervised Learning"

Related tags

Overview

MarginGAN

Owner

Van

Replication Package for "An Empirical Study of the Effectiveness of an Ensemble of Stand-alone Sentiment Detection Tools for Software Engineering Datasets"

Picasso: A CUDA-based Library for Deep Learning over 3D Meshes

NaturalCC is a sequence modeling toolkit that allows researchers and developers to train custom models

Composable transformations of Python+NumPy programsComposable transformations of Python+NumPy programs

Multi-Agent Reinforcement Learning (MARL) method to learn scalable control polices for multi-agent target tracking.

6D Grasping Policy for Point Clouds

Universal Adversarial Examples in Remote Sensing: Methodology and Benchmark

A whale detector design for the Kaggle whale-detector challenge!

Real-Time Social Distance Monitoring tool using Computer Vision

Open standard for machine learning interoperability

Face Depixelizer based on "PULSE: Self-Supervised Photo Upsampling via Latent Space Exploration of Generative Models" repository.

NAS-HPO-Bench-II is the first benchmark dataset for joint optimization of CNN and training HPs.

Pytorch implementation of "Geometrically Adaptive Dictionary Attack on Face Recognition" (WACV 2022)

Code to reproduce results from the paper "AmbientGAN: Generative models from lossy measurements"

[NeurIPS 2021] “Improving Contrastive Learning on Imbalanced Data via Open-World Sampling”,

Automates Machine Learning Pipeline with Feature Engineering and Hyper-Parameters Tuning :rocket:

A spatial genome aligner for analyzing multiplexed DNA-FISH imaging data.

Official Pytorch implementation of the paper: "Locally Shifted Attention With Early Global Integration"

QuickAI is a Python library that makes it extremely easy to experiment with state-of-the-art Machine Learning models.