Data to Actionable Knowledge (DtAK) Lab's repositories
adversarial-robustness-public
Code for AAAI 2018 accepted paper: "Improving the Adversarial Robustness and Interpretability of Deep Neural Networks by Regularizing their Input Gradients"
mbrl-smdp-ode
PyTorch implementation of "Model-based Reinforcement Learning for Semi-Markov Decision Processes with Neural ODEs", NeurIPS 2020
ocbnn-public
General purpose library for BNNs, and implementation of OC-BNNs in our 2020 NeurIPS paper.
prediction-constrained-topic-models
Public repo containing code to train, visualize, and evaluate semi-supervised topic models and baselines for regression/classification on labeled bag-of-words dataset, as described in Hughes et al. AISTATS 2018
POPCORN-POMDP
Implementation of "POPCORN: Partially Observed Prediction Constrained Reinforcement Learning" (Futoma, Hughes, Doshi-Velez, AISTATS 2020)
interactive-reconstruction
Code for Evaluating the Interpretability of Generative Models by Interactive Reconstruction
addressing-leakage
Code for the paper 'Addressing leakage in Concept Bottleneck Models'
hierarchical-disentanglement
Code for Benchmarks, Algorithms, and Metrics for Hierarchical Disentanglement
local-independence-public
Code/figures in Learning Qualitatively Diverse and Interpretable Rules for Classification
optimal-summaries-public
Code repository for the MLHC 2022 paper "Learning Optimal Summaries of Clinical Time-series with Concept Bottleneck Models"
rethinking_discount_reg_public
Simulations for the paper: "Rethinking Discount Regularization: New Interpretations, Unintended Consequences, and Solutions for Regularization in Reinforcement Learning"
anchor-box
This repository contains the code for out work, Guarantee Regions for Local Explanations
kernel_mismatch_workshop
Code for Implications of Gaussian process kernel mismatch for out-of-distribution data (ICML 2023 workshops)
tensorpack
A Neural Net Training Interface on TensorFlow, with focus on speed + flexibility