interpretable models
Interpretable models are AI systems designed to provide clear insights into their inner workings and decision-making processes. Enhancing interpretability is key to understanding model behavior, validating predictions, and ensuring accountability, particularly in high-stakes applications.
- Causal Climate Emulation with Bayesian Filtering
- Causally Reliable Concept Bottleneck Models
- DermaCon-IN: A Multiconcept-Annotated Dermatological Image Dataset of Indian Skin Disorders for Clinical AI Research
- From Black-box to Causal-box: Towards Building More Interpretable Models
- Graph-based Symbolic Regression with Invariance and Constraint Encoding
- Interpretable Next-token Prediction via the Generalized Induction Head
- Interpretable and Parameter Efficient Graph Neural Additive Models with Random Fourier Features
- MANGO: Multimodal Attention-based Normalizing Flow Approach to Fusion Learning
- Object-Centric Concept-Bottlenecks
- PSI: A Benchmark for Human Interpretation and Response in Traffic Interactions
- SPARTAN: A Sparse Transformer World Model Attending to What Matters
- Structured Temporal Causality for Interpretable Multivariate Time Series Anomaly Detection
- Vision Function Layer in Multimodal LLMs