My personal website.

Trustworthy AI/ML course by Professor Birhanu Eshete, University of Michigan, Dearborn.

This repo contains the codes, figures and datasets for the paper - U-Trustworthy Models. Reliability, Competence, and Confidence in Decision-Making.

DSPLab@UMich-Dearborn Website

Birhanu Eshete is an Associate Professor of Computer Science at the University of Michigan, Dearborn. His main research focus is in trustworthy machine learning with emphasis on security, safety, privacy, interpretability, fairness, and the dynamics thereof. He also studies online cybercrime and advanced and persistent threats (APTs).

Explainable Debugger for Black-box Machine Learning Models

KDD 2023 tutorial "Trustworthy Transfer Learning: Transferability and Trustworthiness"

Explanation-guided boosting of machine learning evasion attacks.

Welcome to my Machine Learning repository, where you can find learning materials both from my studies and from various online courses.

Official implementation of NeurIPS 2023 paper "Trade-off Between Efficiency and Consistency for Removal-based Explanations" (https://arxiv.org/abs/2210.17426)

In the dynamic landscape of medical artificial intelligence, this study explores the vulnerabilities of the Pathology Language-Image Pretraining (PLIP) model, a Vision Language Foundation model, under targeted attacks like PGD adversarial attack.

TRIAGE: Characterizing and auditing training data for improved regression (NeurIPS 2023)

Code for the paper "Approximating full conformal prediction at scale via influence functions""

Data-SUITE: Data-centric identification of in-distribution incongruous examples (ICML 2022)

Code from PLDI '21 paper "Provable Repair of Deep Neural Networks."

A School for All Seasons on Trustworthy Machine Learning

Repository for the NeurIPS 2023 paper "Beyond Confidence: Reliable Models Should Also Consider Atypicality"

Morphence: An implementation of a moving target defense against adversarial example attacks demonstrated for image classification models trained on MNIST and CIFAR10.

A project to train your model from scratch or fine-tune a pretrained model using the losses provided in this library to improve out-of-distribution detection and uncertainty estimation performances. Calibrate your model to produce enhanced uncertainty estimations. Detect out-of-distribution data using the defined score type and threshold.

MERLIN is a global, model-agnostic, contrastive explainer for any tabular or text classifier. It provides contrastive explanations of how the behaviour of two machine learning models differs.