Posters

AI systems acquire bias during training, impacting accuracy across sub-populations. This research unveils bias’s dynamic nature, revealing how classifier preferences shift over time, influenced by dat…

This paper presents efficient algorithms to detect whether an agent updates beliefs optimally (Bayesian) or exhibits bias towards their prior beliefs, using information design and signaling schemes.

LLMs’ iterative interactions amplify subtle biases; this paper uses a Bayesian Iterated Learning framework to explain this phenomenon and offers strategies to guide LLM evolution.

Vision transformers surprisingly struggle with visual relations; this study reveals ViTs use distinct perceptual and relational processing stages to solve same/different tasks, highlighting a previous…

Lifelong learning in linear bandits gets a boost! A new algorithm, BOSS, achieves low regret without the usual ‘task diversity’ assumption, opening doors for more realistic sequential multi-task lear…

Researchers drastically sped up high-fidelity deep learning model extraction, improving efficiency by up to 14.8x and challenging previous assumptions on the extraction bottleneck.

AI struggles with collaborating effectively with humans due to unpredictable human behavior. This paper introduces Collaborative Bayesian Policy Reuse (CBPR), a novel framework that leverages meta-ta…

InfoMGF, a novel framework, tackles the limitations of unsupervised multiplex graph learning by refining graph structures, maximizing task-relevant information (both shared and unique), and achieving …

This paper presents a novel, UCB-like algorithm for bandits with stochastic and adversarial constraints, achieving optimal performance without the stringent assumptions of prior primal-dual methods.

Beyond Euclidean spaces, Dual-Space Representation Learning (DSRL) enhances weakly supervised video violence detection by cleverly integrating Euclidean and hyperbolic geometries for superior discrimi…

MolPeg, a novel molecular data pruning framework, enhances model generalization in transfer learning by using a source-free approach and consistently outperforming other methods, even surpassing full-…

This paper presents a novel method to make black box neural networks intervenable using only a small validation set with concept labels, improving the effectiveness of concept-based interventions.

CPDTrack: Human-like Visual Search Boosts Object Tracking!

This research introduces a novel two-phase approach to improve AI model trustworthiness by ensuring both correct predictions and correct rationales. A new dataset with structured rationales and a rat…

Researchers developed AdvRM, a new adversarial patch attack against monocular depth estimation models, which effectively camouflages patches as road markings to mislead depth predictions for any obsta…

BetterDepth: A plug-and-play diffusion refiner boosts zero-shot monocular depth estimation by adding fine details while preserving accurate geometry.

Strong pre-tuned MLPs and meta-tuned default parameters for GBDTs and MLPs improve tabular data classification and regression.

Masked language models can perform in-context learning, challenging the dominance of causal models in this area.

BEND-VLM: A novel, efficient test-time debiasing method for vision-language models, resolving bias without retraining.

O-BELM, a novel diffusion model sampler, achieves mathematically exact inversion with superior sampling quality, offering a new gold standard for diffusion model applications.