Inherent Interpretability

Inherent interpretability in machine learning focuses on designing models and methods that are inherently transparent and understandable, aiming to reduce the "black box" nature of many AI systems. Current research emphasizes developing intrinsically interpretable model architectures, such as those based on decision trees, rule-based systems, and specific neural network designs (e.g., Kolmogorov-Arnold Networks), alongside techniques like feature attribution and visualization methods to enhance understanding of model behavior. This pursuit is crucial for building trust in AI, particularly in high-stakes applications like healthcare and finance, where understanding model decisions is paramount for responsible deployment and effective human-AI collaboration.

Papers

February 7, 2023

Linear-scaling kernels for protein sequences and small molecules outperform deep learning while providing uncertainty quantitation and improved interpretability
Jonathan Parkinson, Wei Wang
Deep Learning Inherent Interpretability Gaussian Process Protein Sequence Small Molecule Linear Kernel Kernel Convolution

February 3, 2023

Improving Interpretability via Explicit Word Interaction Graph Layer
Arshdeep Sekhon, Hanjie Chen, Aman Shrivastava, Zhe Wang, Yangfeng Ji, Yanjun Qi
Natural Language Inherent Interpretability NLP Model Word Usage Graph Natural Language Processing Classifier

January 30, 2023

NeSyFOLD: Neurosymbolic Framework for Interpretable Image Classification
Parth Padalkar, Huaduo Wang, Gopal Gupta
Deep Learning Model Inherent Interpretability Image Classification Back Propagation CNN Model Interpretable Image

January 29, 2023

Incremental Value and Interpretability of Radiomics Features of Both Lung and Epicardial Adipose Tissue for Detecting the Severity of COVID-19 Infection
Ni Yao, Yanhui Tian, Daniel Gama das Neves, Chen Zhao, Claudio Tinoco Mesquita, Wolney de Andrade Martins, Alair Augusto Sarmet Moreira Damas dos Santos, Yanting Li, Chuang Han, Fubao Zhu, Neng Dai, Weihua Zhou
Inherent Interpretability Radiomics Feature Coronavirus Disease 2019 COVID 19 Severity Epicardial Adipose Tissue

January 22, 2023

Interpretability in Activation Space Analysis of Transformers: A Focused Survey
Soniya Vijayakumar
Transformer Megatron Decepticons Latent Space Inherent Interpretability Comprehensive Survey Attention Layer Self Attention Layer Activation Space Feed Forward Layer

January 21, 2023

Towards Flexibility and Interpretability of Gaussian Process State-Space Model
Zhid Lin, Feng Yin, Juan Maroñas
Inherent Interpretability Gaussian Process State Space Model State Space Scalable Variational Sparse GP

January 18, 2023

On the limits of neural network explainability via descrambling
Shashank Sule, Richard G. Spencer, Wojciech Czaja
Neural Network Deep Learning Inherent Interpretability Inverse Problem Path Breaking Emergence Singular Value Decomposition Signal Representation

January 17, 2023

January 16, 2023

Img2Tab: Automatic Class Relevant Concept Discovery from StyleGAN Features for Explainable Image Classification
Youngjae Song, Sung Kuk Shyn, Kwang-su Kim
Inherent Interpretability StyleGAN Latent Concept Based Concept Discovery Interpretable Image Tabular Data Classification CNN Classifier

January 15, 2023

Generalized Neural Closure Models with Interpretability
Abhinav Gupta, Pierre F. J. Lermusiaux
Inherent Interpretability Differentiable Physic Dynamical Model Fidelity Model Synaptic Delay Closure Model

January 12, 2023

Tracr: Compiled Transformers as a Laboratory for Interpretability
David Lindner, János Kramár, Sebastian Farquhar, Matthew Rahtz, Thomas McGrath, Vladimir Mikulik
Transformer Megatron Decepticons Inherent Interpretability Interpretability Method Decoder Only Transformer Likely LAB of Origin End to End Compiler

January 11, 2023

The Berkelmans-Pries Feature Importance Method: A Generic Measure of Informativeness of Features
Joris Pries, Guus Berkelmans, Sandjai Bhulai, Rob van der Mei
Machine Learning Model Human Prediction Inherent Interpretability Feature Wise Feature Importance

January 5, 2023

DANLIP: Deep Autoregressive Networks for Locally Interpretable Probabilistic Forecasting
Ozan Ozyegen, Juyoung Wang, Mucahit Cevik
Human Prediction Inherent Interpretability Probabilistic Time Series Forecasting Interpretable Forecast Deep Probabilistic Forecasting

January 4, 2023

Iterated Decomposition: Improving Science Q&A by Supervising Reasoning Processes
Justin Reppert, Ben Rachbach, Charlie George, Luke Stebbing, Jungwon Byun, Maggie Appleton, Andreas Stuhlmüller
Language Model Inherent Interpretability Complex Reasoning Reasoning Process Compositional Learning Compositional Approach Stagewise Decomposition

December 30, 2022

On the Interpretability of Attention Networks
Lakshmi Narayan Pandey, Rahul Vashisht, Harish G. Ramaswamy
Inherent Interpretability Attention Mechanism Attention Network CNN Architecture Attention Based Architecture

December 23, 2022

Security and Interpretability in Automotive Systems
Shailja Thakur
Inherent Interpretability Security Related Controller Area Network Automotive Application Context Impersonation

December 21, 2022

Interpretability and causal discovery of the machine learning models to predict the production of CBM wells after hydraulic fracturing
Chao Min, Guoquan Wen, Liangjie Gou, Xiaogang Li, Zhaozhong Yang
Machine Learning Model Inherent Interpretability Causal Discovery Causal Graph Latent Causal Oil Well

December 20, 2022

A Comparison Between Tsetlin Machines and Deep Neural Networks in the Context of Recommendation Systems
Karl Audun Borgersen, Morten Goodwin, Jivitesh Sharma
Deep Neural Network Deep Learning Model Machine Learning Model Inherent Interpretability Context Information Consistent Comparison Recommendation System Recommendation Scenario Tsetlin Machine

December 19, 2022

Continuous Semi-Supervised Nonnegative Matrix Factorization
Michael R. Lindstrom, Xiaofu Ding, Feng Liu, Anand Somayajula, Deanna Needell
Inherent Interpretability Matrix Factorization Non Negative Matrix