Signature Activation: A Sparse Signal View for Holistic Saliency
The adoption of machine learning in healthcare calls for model transparency and explainability. In this work, we introduce Signature Activation, a saliency method that generates holistic and class-agnostic explanations for Convolutional Neural Network (CNN) outputs. Our method exploits the fact that certain kinds of medical images, such as angiograms, have clear foreground and background objects. We give theoretical explanation to justify our methods. We show the potential use of our method in clinical settings through evaluating its efficacy for aiding the detection of lesions in coronary angiograms.
Code (1)
Similar Papers 제목 키워드 기반
Task-Conditioned Routing Signatures in Sparse Mixture-of-Experts Transformers
Sparse Mixture-of-Experts (MoE) architectures enable efficient scaling of large language models through conditional computation, yet the routing mechanisms responsible for expert selection remain poorly understood. In th…
MiA-Signature: Approximating Global Activation for Long-Context Understanding
A growing body of work in cognitive science suggests that reportable conscious access is associated with \emph{global ignition} over distributed memory systems, while such activation is only partially accessible as indiv…
Long-Context UnderstandingLLM Self-Recognition: Steering and Retrieving Activation Signatures
Recent advances in interpretability suggest that large language models (LLMs) implicitly encode signals in their generated text that enable self-recognition of their outputs. We demonstrate that this capability is reliab…
Sparse maximal update parameterization: A holistic approach to sparse training dynamics
Several challenges make it difficult for sparse neural networks to compete with dense models. First, setting a large fraction of weights to zero impairs forward and gradient signal propagation. Second, sparse studies oft…
Language ModelingLanguage ModellingPre-Intervention Prediction of Sparse Autoencoder Steering Side Effects
Sparse autoencoder (SAE) features are increasingly used to steer language models, but feature steering is rarely clean: the same intervention can behave inconsistently across contexts and perturb unrelated features. We i…