The code generates simulated research landscapes of any given bias. It can be used to calculate the Machine Learning Gain under different conditions.
-
Updated
Jun 22, 2022 - Python
The code generates simulated research landscapes of any given bias. It can be used to calculate the Machine Learning Gain under different conditions.
A teacher model with minimal bias or calibrate its outputs to reduce bias transfer during distillation. Focus on distilling information from teacher features that are less prone to bias, like semantic representations instead of raw activations.
This German scientific poster (developed as HTML5/React web app) shows the potential implications of racial/social/... Biases in the context of Machine Learning
All the material needed to implement FairVIC for bias mitigation in deep neural networks.
A demonstration of detecting and mitigating bias in AI.
Explore how bias can be detected and mitigated in real-world datasets
A small and simple prototype designed to alert users of the bias of the news source.
Code for the paper The Other Side of Compression: Measuring Bias in Pruned Transformers (IDA23)
Codebase for political bias mitigation in ChatGPT using Anthropic's ConstitutionalAI principles
Coursework solutions for a 2nd year Computer Science sub-module on Bias in AI @ Durham University. Attempts to minimise gender bias when using ML to predict income.
Pytorch implementation of 'Explaining text classifiers with counterfactual representations'
An ML competition on CodaLab to estimate the age from images while mitigating the bias
Detecting & Mitigating Bias: Gender, Age & Site
Explainable AI & fashion talk & experiments
Research POC on the mitigation of bias in large language models (FLAN-T5 and Bloomz) through fine-tuning.
Gender classification model that uses a CNN to classify images of faces as male or female. The notebook includes code for data preprocessing, model architecture, training, and evaluation which will then be used for algorithmic bias detection.
Mitigating algorithmic bias using regularisation and reweighing with bi-objective (accuracy and fairness) measure
Fairness in Face Presentation Attack Detection
Data and Results for WWW paper
Add a description, image, and links to the bias-mitigation topic page so that developers can more easily learn about it.
To associate your repository with the bias-mitigation topic, visit your repo's landing page and select "manage topics."