Reactive Model Correction: Mitigating Harm to Task-Relevant Features via Conditional Bias Suppression
Abstract Deep Neural Networks are prone to learning and relying on spurious correlations in
the training data which for high-risk applications can have fatal consequences. Various …
the training data which for high-risk applications can have fatal consequences. Various …
Explain to Question not to Justify
Explainable Artificial Intelligence (XAI) is a young but very promising field of research.
Unfortunately, the progress in this field is currently slowed down by divergent and …
Unfortunately, the progress in this field is currently slowed down by divergent and …
Neural Concept Binder
The challenge in object-based visual reasoning lies in generating descriptive yet distinct
concept representations. Moreover, doing this in an unsupervised fashion requires human …
concept representations. Moreover, doing this in an unsupervised fashion requires human …
FI-CBL: A Probabilistic Method for Concept-Based Learning with Expert Rules
LV Utkin, AV Konstantinov, SR Kirpichenko - arXiv preprint arXiv …, 2024 - arxiv.org
A method for solving concept-based learning (CBL) problem is proposed. The main idea
behind the method is to divide each concept-annotated image into patches, to transform the …
behind the method is to divide each concept-annotated image into patches, to transform the …
Incorporating Expert Rules into Neural Networks in the Framework of Concept-Based Learning
AV Konstantinov, LV Utkin - arXiv preprint arXiv:2402.14726, 2024 - arxiv.org
A problem of incorporating the expert rules into machine learning models for extending the
concept-based learning is formulated in the paper. It is proposed how to combine logical …
concept-based learning is formulated in the paper. It is proposed how to combine logical …
PURE: Turning Polysemantic Neurons Into Pure Features by Identifying Relevant Circuits
The field of mechanistic interpretability aims to study the role of individual neurons in Deep
Neural Networks. Single neurons, however, have the capability to act polysemantically and …
Neural Networks. Single neurons, however, have the capability to act polysemantically and …
Position: Explain to Question not to Justify
Explainable Artificial Intelligence (XAI) is a young but very promising field of research.
Unfortunately, the progress in this field is currently slowed down by divergent and …
Unfortunately, the progress in this field is currently slowed down by divergent and …