niftynans Goto Github PK

algo_stability_nns_ifs

bowlership

Team sports are all about player synergies; deciding when to let your partner take the shot in Tennis doubles or anticipating the next pass in Football. This is why often the best single’s champions are not necessarily the best doubles’ champions in Tennis, and the team with more star players does not necessarily win the Football match. In this paper, we investigate bowler pairs. We introduce and precisely define a bowling partnership. Bowlers bowl in pairs from opposite ends of the field alternately. It is often the case that a pair of bowlers bowl several consecutive overs alternately. We define such pairs of bowlers a bowling pair. Are some bowling pairs more effective than others? Are they effective in terms of saving runs or taking wickets?

cfactual_causal

This project focuses on counterfactual generation for causal inference and model evaluation using the archipelago penguin specie classification dataset.

cfr_covariate_matching

CFR

climax

faster-lime-ifs

LIME but faster, precise, consistent, and robust

fire_evac_maze

This project aims to provide the best path for evacuation of multiple agents on a given floor plan for multiple fire origination points through a maze solver algorithm which uses the Dijkstra Algorithm.

fish_cfrnet

Official implementation of paper Gradient Matching for Domain Generalization

grad_cam_plus_plus

A generalized gradient-based CNN visualization technique

hsic-attribution-method

hsic_attribution_method

if_lime-experiments

Code for all experiments.

if_unravel

Influence Based Acquisitions for Post-Hoc Explainability

influence-release

influence_subsampling_lime

Official Implementation of Unweighted Data Subsampling via Influence Function - AAAI 2020

invariance-via-subsampling

source code for 'Finding Valid Adjustments under Non-ignorability with Minimal DAG Knowledge' by A. Shah, K. Shanmugam, K. Ahuja

lime_experiments_ifs

Experiments with LIME

lime_stability_ifs

modeling-uncertainty-local-explainability-climax

niftynans

Config files for my GitHub profile.

nn-explainer

sc-cfr_matched

Codes and datasets for reproducing results of AAMAS 2020.

topo

Optimizing NOTEARS Objectives via Topological Swaps

unweighted-data-subsampling-via-influence-function