Dataloader for CIFAR-N

CIFAR-10N

noise_label = torch.load('./data/CIFAR-10_human.pt')
clean_label = noise_label['clean_label']
worst_label = noise_label['worse_label']
aggre_label = noise_label['aggre_label']
random_label1 = noise_label['random_label1']
random_label2 = noise_label['random_label2']
random_label3 = noise_label['random_label3']

CIFAR-100N

noise_label = torch.load('./data/CIFAR-100_human.pt')
clean_label = noise_label['clean_label']
noisy_label = noise_label['noisy_label']

Training on CIFAR-N with the Cross-Entropy loss

CIFAR-10N

# NOISE_TYPE: [clean, aggre, worst, rand1, rand2, rand3]
# Use human annotations
CUDA_VISIBLE_DEVICES=0 python3 main.py --dataset cifar10 --noise_type NOISE_TYPE --is_human
# Use the synthetic noise that has the same noise transition matrix as human annotations
CUDA_VISIBLE_DEVICES=0 python3 main.py --dataset cifar10 --noise_type NOISE_TYPE

CIFAR-100N

# NOISE_TYPE: [clean100, noisy100]
# Use human annotations
CUDA_VISIBLE_DEVICES=0 python3 main.py --dataset cifar100 --noise_type NOISE_TYPE --is_human
# Use the synthetic noise that has the same noise transition matrix as human annotations
CUDA_VISIBLE_DEVICES=0 python3 main.py --dataset cifar100 --noise_type NOISE_TYPE

Additional dataset information

We include additional side information during the noisy-label collection in side_info_cifar10N.csv and side_info_cifar100N.csv. A brief introduction of these two files:

Image-batch: a subset of indexes of the CIFAR training images.
Worker-id: the encrypted worker id on Amazon Mechanical Turk.
Work-time-in-seconds: the time (in seconds) a worker spent on annotating the corresponding image batch.

Human annotated noisy labels for CIFAR-10 and CIFAR-100.

Related tags

Overview

Dataloader for CIFAR-N

CIFAR-10N

CIFAR-100N

Training on CIFAR-N with the Cross-Entropy loss

CIFAR-10N

CIFAR-100N

Additional dataset information

Owner

[email protected]

Source code of CIKM2021 Long Paper "PSSL: Self-supervised Learning for Personalized Search with Contrastive Sampling".

Implementation supporting the ICCV 2017 paper "GANs for Biological Image Synthesis"

ML From Scratch

Riemannian Convex Potential Maps

AnimationKit: AI Upscaling & Interpolation using Real-ESRGAN+RIFE

Intro-to-dl - Resources for "Introduction to Deep Learning" course.

Official PyTorch implementation of MX-Font (Multiple Heads are Better than One: Few-shot Font Generation with Multiple Localized Experts)

Code for Overinterpretation paper Overinterpretation reveals image classification model pathologies

Realtime segmentation with ENet, the fast and accurate segmentation net.

Pretrained Cost Model for Distributed Constraint Optimization Problems

PyTorch implementation of our ICCV 2021 paper, Interpretation of Emergent Communication in Heterogeneous Collaborative Embodied Agents.

rastrainer is a QGIS plugin to training remote sensing semantic segmentation model based on PaddlePaddle.

Third party Pytorch implement of Image Processing Transformer (Pre-Trained Image Processing Transformer arXiv:2012.00364v2)

Code and project page for ICCV 2021 paper "DisUnknown: Distilling Unknown Factors for Disentanglement Learning"

Deep RGB-D Saliency Detection with Depth-Sensitive Attention and Automatic Multi-Modal Fusion (CVPR'2021, Oral)

JumpDiff: Non-parametric estimator for Jump-diffusion processes for Python

A colab notebook for training Stylegan2-ada on colab, transfer learning onto your own dataset.

Minimal fastai code needed for working with pytorch

Lolviz - A simple Python data-structure visualization tool for lists of lists, lists, dictionaries; primarily for use in Jupyter notebooks / presentations

Repository of Jupyter notebook tutorials for teaching the Deep Learning Course at the University of Amsterdam (MSc AI), Fall 2020