csinva
Senior researcher @Microsoft interpreting ML models in science and medicine. PhD from UC Berkeley.
Senior researcherMicrosoft research
Pinned Repositories
csinva.github.io
Slides, paper notes, class notes, blog posts, and research on ML 📉, statistics 📊, and AI 🤖.
gan-vae-pretrained-pytorch
Pretrained GANs + VAEs + classifiers for MNIST/CIFAR in pytorch.
gpt-paper-title-generator
Generating paper titles (and more!) with GPT trained on data scraped from arXiv.
hierarchical-dnn-interpretations
Using / reproducing ACD from the paper "Hierarchical interpretations for neural network predictions" 🧠 (ICLR 2019)
imodels
Interpretable ML package 🔍 for concise, transparent, and accurate predictive modeling (sklearn-compatible).
imodelsX
Scikit-learn friendly library to interpret, and prompt-engineer text datasets using large language models.
iprompt
Finding semantically meaningful and accurate prompts.
tree-prompt
Tree prompting: easy-to-use scikit-learn interface for improved prompting.
deep-explanation-penalization
Code for using CDEP from the paper "Interpretations are useful: penalizing explanations to align neural networks with prior knowledge" https://arxiv.org/abs/1909.13584
covid19-severity-prediction
Extensive and accessible COVID-19 data + forecasting for counties and hospitals. 📈
csinva's Repositories
csinva/gan-vae-pretrained-pytorch
Pretrained GANs + VAEs + classifiers for MNIST/CIFAR in pytorch.
csinva/hierarchical-dnn-interpretations
Using / reproducing ACD from the paper "Hierarchical interpretations for neural network predictions" 🧠 (ICLR 2019)
csinva/disentangled-attribution-curves
Using / reproducing DAC from the paper "Disentangled Attribution Curves for Interpreting Random Forests and Boosted Trees"
csinva/transformation-importance
Using / reproducing TRIM from the paper "Transformation Importance with Applications to Cosmology" 🌌 (ICLR Workshop 2020)
csinva/glaucoma-diagnosis
Code for diagnosing glaucoma from Lumos lens
csinva/news-balancer
News Balancer takes a story and provides articles on that story with credibility and varying political bias. The homepage will randomly generate a story from its archives, but a user can type in a query to get stories relating to their query along with their credibility / political bias.
csinva/dnn-ensemble
Testing the properties of ensembled neural networks.
csinva/max-activation-interpretation-pytorch
Code for creating maximal activation images (like Deep Dream) in pytorch with various regularizations / losses.
csinva/abide-multitask-learning
Multi-task learning of functional connectivity on the ABIDE dataset.
csinva/local-vae
Making locally disentangled vaes.
csinva/neural-spike-sorting
Experimental code for performing spike sorting using a neural network.
csinva/pyfim-clone
Clone of pyfim making it installable as a dependency. Copied from http://www.borgelt.net/pyfim.html
csinva/trees-to-networks
Bridging random forests and deep neural networks. Partial implementation of "Neural Random Forests" https://arxiv.org/abs/1604.07143
csinva/dnn-experiments
A set of scripts and experiments making it easier to analyze deep learning empirically.
csinva/hummingbird-tracking
Code for tracking various things in hummingbird video
csinva/imodels-playground
Demos for visualizing how rule-based models work.
csinva/mouse-brain-decoding
Decoding images from calcium recordings using data from stringer et al. 2018.
csinva/scattering-transform-experiments
Repository for experiments with scattering transforms
csinva/stable-interpretation
Exploring ways to extract stable interpretations from neural networks.
csinva/arxiv-copier
Extension for copying the title + url of an arXiv page via right click
csinva/axon-ap-propagation
Code for simulations of action potential propagation
csinva/global-sports-analysis
Analyzing how different factors influence global sports rankings
csinva/mini-games
Code for simple games made in java + google sheets.
csinva/younet
Learning natural language models based on personalized messages.
csinva/biophysical-threshold-simulations
Simulations for characterizing stochastic threshold in biophysical neurons.
csinva/conv-net-scripts
Scripts for evaluation of convolutional networks
csinva/donation-swapping
A website to facilitate routing opposing donations to charity.
csinva/hpa-interp
Interpretations on the HPA dataset.
csinva/nano-descriptions
Algorithms for machine-learning analysis and descriptions of nanomaterials
csinva/total-variation-abide
Experiments with using total variation regularization on the ABIDE fmri dataset.