Benjamin Wong

Pierre Moreau, Emeline Pineau Ferrand, Yann Choho et al.

Concept Bottleneck Models (CBMs) are interpretable models that route predictions through a layer of human-interpretable concepts. While widely studied in vision and, more recently, in NLP, CBMs remain largely unexplored in multimodal settings. For their explanations to be faithful, CBMs must satisfy two conditions: concepts must be properly detected, and concept representations must encode only their intended semantics, without smuggling extraneous task-relevant or inter-concept information into final predictions, a phenomenon known as leakage. Existing approaches treat concept detection and leakage mitigation as separate problems, and typically improve one at the expense of predictive accuracy. In this work, we introduce f-CBM, a faithful multimodal CBM framework built on a vision-language backbone that jointly targets both aspects through two complementary strategies: a differentiable leakage loss to mitigate leakage, and a Kolmogorov-Arnold Network prediction head that provides sufficient expressiveness to improve concept detection. Experiments demonstrate that f-CBM achieves the best trade-off between task accuracy, concept detection, and leakage reduction, while applying seamlessly to both image and text or text-only datasets, making it versatile across modalities.

Benjamin Wong

The purpose of this study is to test the effectiveness of current straggler mitigation techniques over different important iterative convergent machine learning(ML) algorithm including Matrix Factorization (MF), Multinomial Logistic Regression (MLR), and Latent Dirichlet Allocation (LDA) . The experiment was conducted to implemented using the FlexPS system, which is the latest system implementation that employ parameter server architecture. The experiment employed the Bulk Synchronous Parallel (BSP) computational model to examine the straggler problem in Parameter Server on Iterative Convergent Distributed Machine Learning. Moreover, the current research analyzes the experimental arrangement of the parameter server strategy concerning the parallel learning problems by injecting universal straggler patterns and executing latest mitigation techniques. The findings of the study are significant in that as they will provide the necessary platform for conducting further research into the problem and allow the researcher to compare different methods for various applications. The outcome is therefore expected to facilitate the development of new techniques coupled with new perspectives in addressing this problem.

Arthur Vogels, Benjamin Wong, Yann Choho et al.

Activation steering methods control large language model (LLM) behavior by modifying internal activations at inference time. However, most existing activation steering methods rely on a fixed steering strength, leading to either insufficient control or unadapted intervention that degrades text plausibility and coherence. We introduce In-Distribution Steering (IDS), a novel method that adapts steering strength based on the input data distribution in representation space. IDS dynamically adjusts interventions according to how far a given input lies within the distribution, enabling adaptive intervention and generation stability during text generation. Experiments demonstrate that IDS achieves strong accuracy on classification tasks while producing coherent text without collapse, making IDS particularly well suited for real-world applications.