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

November 12, 2022

Deep Reinforcement Learning with Vector Quantized Encoding
Liang Zhang, Justin Lieffers, Adarsh Pyarelal
Reinforcement Learning Deep Reinforcement Learning Inherent Interpretability Vector Quantization

November 11, 2022

Rethinking Log Odds: Linear Probability Modelling and Expert Advice in Interpretable Machine Learning
Danial Dervovic, Nicolas Marchesotti, Freddy Lecue, Daniele Magazzeni
Inherent Interpretability Interpretable Machine Learning Additive Model Additive Intervention

November 10, 2022

What Makes a Good Explanation?: A Harmonized View of Properties of Explanations
Zixi Chen, Varshini Subhash, Marton Havasi, Weiwei Pan, Finale Doshi-Velez
Machine Learning Inherent Interpretability Line by Line Explanation Interpretable Machine Learning Many Property Unified View Interpretable Machine Quality Explanation Interpretable Machine Learning Method

November 2, 2022

On the Safety of Interpretable Machine Learning: A Maximum Deviation Approach
Dennis Wei, Rahul Nair, Amit Dhurandhar, Kush R. Varshney, Elizabeth M. Daly, Moninder Singh
Inherent Interpretability Human SAFETY Interpretable Machine Learning Explainable Machine Learning Least Absolute Deviation

November 1, 2022

Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 small
Kevin Wang, Alexandre Variengien, Arthur Conmy, Buck Shlegeris, Jacob Steinhardt
Language Model Inherent Interpretability Wild Challenge Mechanistic Interpretability Interpretability Method GPT 2 Biological Circuit Mechanistic Understanding

October 31, 2022

October 29, 2022

Interpretable CNN-Multilevel Attention Transformer for Rapid Recognition of Pneumonia from Chest X-Ray Images
Shengchao Chen, Sufen Ren, Guanjun Wang, Mengxing Huang, Chenyang Xue
Inherent Interpretability Chest X Ray Chest X Ray Image Ventilator Associated Pneumonia Pneumonia Detection

October 28, 2022

Towards trustworthy multi-modal motion prediction: Holistic evaluation and interpretability of outputs
Sandra Carrasco Limeros, Sylwia Majchrowska, Joakim Johnander, Christoffer Petersson, Miguel Ángel Sotelo, David Fernández Llorca
Autonomous Vehicle Inherent Interpretability Motion Prediction Structured Output Holistic Evaluation

October 15, 2022

Decoupling Deep Learning for Interpretable Image Recognition
Yitao Peng, Yihang Liu, Longzhen Yang, Lianghua He
Deep Learning Inherent Interpretability Prototypical Network Prototype Network Interpretable Image Recognition

October 13, 2022

Interpreting Neural Policies with Disentangled Tree Representations
Tsun-Hsuan Wang, Wei Xiao, Tim Seyde, Ramin Hasani, Daniela Rus
Inherent Interpretability Robot Learning Disentanglement Capability Disentangled Representation Neural Dynamic Neural Network Policy Compact Neural Network

October 12, 2022

That's the Wrong Lung! Evaluating and Improving the Interpretability of Unsupervised Multimodal Encoders for Medical Data
Denis Jered McInerney, Geoffrey Young, Jan-Willem van de Meent, Byron C. Wallace
Inherent Interpretability Multimodal Model Human Annotation Medical Data Multimodal Capability Multimodal Encoder Soft Alignment

October 9, 2022

A Detailed Study of Interpretability of Deep Neural Network based Top Taggers
Ayush Khot, Mark S. Neubauer, Avik Roy
Deep Neural Network Inherent Interpretability Explainable AI High Explainability DNN Model Comprehensive Study Po Tagger Particle Flow

October 7, 2022

1st ICLR International Workshop on Privacy, Accountability, Interpretability, Robustness, Reasoning on Structured Data (PAIR^2Struct)
Hao Wang, Wanyu Lin, Hao He, Di Wang, Chengzhi Mao, Muhan Zhang
Artificial Intelligence Native Robustness Inherent Interpretability Complex Reasoning Graph Structured Data Structured Data Structured Knowledge

October 4, 2022

Polysemanticity and Capacity in Neural Networks
Adam Scherlis, Kshitij Sachan, Adam S. Jermyn, Joe Benton, Buck Shlegeris
Neural Network Inherent Interpretability Individual Neuron Capacity Loss Polysemous Word

October 3, 2022

Enhancing Interpretability and Interactivity in Robot Manipulation: A Neurosymbolic Approach
Georgios Tziafas, Hamidreza Kasaei
Inherent Interpretability Visual Grounding Robot Manipulation Interactive No Code 3D Vision Language Neurosymbolic Approach Neurosymbolic Framework Visual Language Reasoning

October 2, 2022

What shapes the loss landscape of self-supervised learning?
Liu Ziyin, Ekdeep Singh Lubana, Masahito Ueda, Hidenori Tanaka
Native Robustness Self Supervised Learning Inherent Interpretability Loss Landscape Dimensional Collapse

September 30, 2022

Evaluation of importance estimators in deep learning classifiers for Computed Tomography
Lennart Brocki, Wistan Marchadour, Jonas Maison, Bogdan Badic, Panagiotis Papadimitroulas, Mathieu Hatt, Franck Vermet, Neo Christopher Chung
Global Evaluation Inherent Interpretability Deep Convolutional Neural Network Saliency Map Computed Tomography Deep Learning Classifier Importance Estimation

September 25, 2022

A Deep Learning Approach to Analyzing Continuous-Time Systems
Cory Shain, William Schuler
Deep Learning Inherent Interpretability Deep Learning Approach Complex Process Continuous Time

Inherent Interpretability

Papers

Deep Reinforcement Learning with Vector Quantized Encoding

Rethinking Log Odds: Linear Probability Modelling and Expert Advice in Interpretable Machine Learning

What Makes a Good Explanation?: A Harmonized View of Properties of Explanations

On the Safety of Interpretable Machine Learning: A Maximum Deviation Approach

Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 small

What is my math transformer doing? -- Three results on interpretability and generalization

Trade-off Between Efficiency and Consistency for Removal-based Explanations

SoK: Modeling Explainability in Security Analytics for Interpretability, Trustworthiness, and Usability

Interpretable CNN-Multilevel Attention Transformer for Rapid Recognition of Pneumonia from Chest X-Ray Images

Towards trustworthy multi-modal motion prediction: Holistic evaluation and interpretability of outputs

Decoupling Deep Learning for Interpretable Image Recognition

Interpreting Neural Policies with Disentangled Tree Representations

That's the Wrong Lung! Evaluating and Improving the Interpretability of Unsupervised Multimodal Encoders for Medical Data

A Detailed Study of Interpretability of Deep Neural Network based Top Taggers

1st ICLR International Workshop on Privacy, Accountability, Interpretability, Robustness, Reasoning on Structured Data (PAIR^2Struct)

Polysemanticity and Capacity in Neural Networks

Enhancing Interpretability and Interactivity in Robot Manipulation: A Neurosymbolic Approach

What shapes the loss landscape of self-supervised learning?

Evaluation of importance estimators in deep learning classifiers for Computed Tomography

A Deep Learning Approach to Analyzing Continuous-Time Systems