Concept Bottleneck Models
PulseAugur coverage of Concept Bottleneck Models — every cluster mentioning Concept Bottleneck Models across labs, papers, and developer communities, ranked by signal.
7 day(s) with sentiment data
-
New research tackles VLM hallucinations, distillation, and interpretability
Researchers are developing new methods to improve the capabilities and reliability of vision-language models (VLMs). One approach, DCLA, focuses on mitigating hallucinations by ensuring consistency across different laye…
-
New NeRD framework boosts AI interpretability in medical diagnosis
Researchers have introduced NeRD (Neuro-Symbolic Rule Distillation), a novel framework designed to enhance interpretability and efficiency in medical image diagnosis. NeRD addresses limitations in existing methods by ge…
-
New AI Models Enhance Interpretability and Reliability in Deep Learning · 4 sources tracked
Researchers have introduced Multimodal Concept Bottleneck Models (MM-CBMs) to enhance the interpretability of deep learning by aligning image and text embeddings with natural concepts. This new approach aims to overcome…
-
New framework enhances 3D generative model interpretability
Researchers have developed a framework called 3D-CBM to enhance interpretability in 3D generative models by integrating Concept Bottleneck Models. This approach aims to bridge the semantic gap in deep geometric learning…
-
AI concept learning unified by geometric framework
Researchers have developed a geometric framework that unifies supervised and unsupervised concept learning in AI models. This approach views both Concept Bottleneck Models (CBMs) and Sparse Autoencoders (SAEs) as learni…
-
New Causal Neural Probabilistic Circuit Enhances Model Interpretability
Researchers have developed a new model called the Causal Neural Probabilistic Circuit (CNPC) to improve the interpretability and intervention capabilities of Concept Bottleneck Models (CBMs). Unlike traditional CBMs tha…
-
New framework boosts interpretable medical image diagnosis
Researchers have developed a new semi-supervised framework for medical image diagnosis that enhances interpretability and efficiency. This approach utilizes dual-level hypergraph learning to model complex relationships …
-
New benchmarks and methods enhance AI interpretability with concept bottleneck models
Researchers are developing new benchmarks and methods for concept bottleneck models (CBMs), which aim to make AI decisions more interpretable by using high-level concepts. One paper introduces synthetic benchmarks to ev…
-
New framework enhances AI model interpretability with multiple concept experts
Researchers have introduced a new framework called Mixture of Concept Bottleneck Experts (M-CBE) to enhance the interpretability and accuracy of concept bottleneck models. This framework allows for the use of multiple p…
-
New Method Uncovers Interpretable Error Slices in Deep Learning Models
Researchers have developed CB-SLICE, a novel method for discovering interpretable error slices in deep learning models. This approach leverages Concept Bottleneck Models (CBMs) to directly link model failures to human-u…
-
New AI Model Enhances Trustworthy Open-Ended Grading in Education
Researchers have developed REC-CBM, a novel concept bottleneck model designed for trustworthy open-ended grading in educational settings. This model addresses limitations in existing systems by explicitly incorporating …
-
New Matryoshka Models Enhance AI Interpretability and Efficiency
Researchers have introduced Matryoshka Concept Bottleneck Models (MCBMs), a novel architecture designed to improve the interpretability and efficiency of deep learning models. MCBMs organize concepts hierarchically, all…
-
New CBM vulnerability exposes interpretable AI to adversarial attacks
Researchers have identified a new vulnerability in Concept Bottleneck Models (CBMs), a type of interpretable machine learning architecture. The study reveals that manipulating the explicit concept activations within CBM…
-
New AI models ground concepts in visual prototypes for better interpretability
Researchers have developed Prototype-Grounded Concept Models (PGCMs) to enhance the interpretability of deep learning models. Unlike previous Concept Bottleneck Models, PGCMs ground concepts in visual prototypes, allowi…
-
New CLIF method enhances NLP model interpretability with concept-level influence functions
Researchers have developed CLIF, a new method using influence functions to improve the interpretability of NLP models. This approach can identify influential training data points, both beneficial and detrimental, and al…
-
Vision models' metonymy undermines attention-based interpretability, study finds
A new research paper published on arXiv introduces the concept of "visual metonymy" in vision models, where parts of an object encode information about the whole object. This phenomenon undermines the interpretability o…
-
GraphCBMs enhance AI interpretability by modeling concept relationships
Researchers have introduced Graph Concept Bottleneck Models (GraphCBMs) to address limitations in existing Concept Bottleneck Models (CBMs). Traditional CBMs assume concepts are independent, ignoring their inherent corr…
-
New CREDENCE framework decomposes AI concept uncertainty for better decision-making
Researchers have developed CREDENCE, a new framework for Credal Concept Bottleneck Models (CBMs) that effectively separates epistemic and aleatoric uncertainty in predictions. This decomposition allows for more nuanced …
-
Researchers identify concept inconsistency in dermoscopic models, impacting accuracy.
Researchers have identified significant concept-level inconsistencies within the Derm7pt dermoscopy dataset, which limit the accuracy of Concept Bottleneck Models (CBMs). By applying rough set theory, they found that 16…