Tag
This paper presents the first implementation of an infra-Bayesian reinforcement learning agent, demonstrating that it outperforms classical RL in worst-case regret and handles Newcomb's problem optimally, offering a step toward robustness under model misspecification.
This paper introduces ASAM (Adversarial Subspace Alignment for Multimodal Knowledge Editing), a method that improves the generality of intrinsic multimodal knowledge editing by generating adversarial yet semantically coherent variants and enforcing low-rank alignment of representations, addressing the limited generalization of existing methods.
Introduces GARD, a diffusion-based framework that operates in the feature space of a feed-forward 3D reconstructor to jointly recover scene geometry and high-quality imagery from degraded inputs.
This paper investigates the conflict between instruction-following and pattern completion in LLMs, finding that instruction-following is brittle under induction pressure and varies widely across models, with output diversity being the primary factor for robustness.
This paper proposes a robust subspace-constrained quadratic model for learning low-dimensional structures from high-dimensional data, accommodating heavy-tailed noise. A gradient-based algorithm with backtracking line search is developed, and experiments show improved robustness and reconstruction accuracy.
SpaceDG is a large-scale dataset and benchmark that evaluates multimodal language models' spatial reasoning robustness under visual degradations like motion blur and low light, revealing significant performance gaps and showing that fine-tuning on SpaceDG improves robustness without degrading clean image performance.
A survey on quantum adversarial machine learning, covering attacks, defenses, and theoretical underpinnings.
This paper studies whether tabular foundation models based on pretrained prior-data fitted networks (PFNs) can generalize to strategic tabular data where individuals modify features after deployment. It proposes Strategic Prior-data Fitted Network (SPN), an inference-time framework that aligns PFN predictions with the post-manipulation distribution without retraining.
This paper identifies Footprint Bias in document layout analysis robustness evaluation and proposes a structure-aware auditing framework that decouples probe construction and pathway attribution, showing that small structurally targeted probes cause comparable downstream degradation to larger perturbations.
This paper introduces SLIM, a minimal architecture that decouples communication from policy representation in multi-agent reinforcement learning, achieving state-of-the-art performance under bandwidth constraints with minimal degradation.
This paper studies adversarial action masking in self-play reinforcement learning, where an attacker selectively removes legal actions from a victim's action set. The attack is shown to be significantly more damaging than random masking or perturbation baselines across multiple environments and algorithms, and victims do not recover under extended training.
MuteBench is a benchmark for evaluating multimodal fusion models under modality missing and within-modality missing conditions across clinical datasets. It provides insights into architecture robustness and suggests that diffusion-based imputation can help.
This paper empirically studies LLMs' legal reasoning in tax law, showing that data contamination inflates performance and that neuro-symbolic hybrid systems offer more reliable and robust generalization than monolithic LLMs.
This paper introduces an Information Bottleneck Adapter (IB-Adapter) for Vision-Language-Action (VLA) models to improve robustness against unseen visual disturbances without requiring extra data, achieving up to 30% improvement with minimal parameter overhead.
This paper investigates the resilience of AI-generated text detection methods (fine-tuned RoBERTa, Binoculars, text feature analysis, and ensembles) against paraphrasing attacks, finding that Binoculars-inclusive ensembles are most effective but also most vulnerable to attacks, highlighting a dichotomy between performance and resilience.
This paper proposes a novel transformer verification approach that uses ReLU to represent precise but non-linear bounds for dot products, enabling precise and efficient verification. The method outperforms state-of-the-art baselines on sentiment analysis models.
This paper evaluates six open-weight LLMs on biomedical QA under conflicting evidence conditions, revealing accuracy drops and prediction flips, and proposes a conflict-aware abstention score that improves selective accuracy.
This paper proposes a method for improving reasoning in large language models by recoding data to explicitly represent relationships, enabling efficient principled reasoning with polynomial-time learnability for relational rules, which addresses hallucinations and supports sound reasoning across multiple calls.
SAE-FT introduces a novel fine-tuning method for CLIP models that uses sparse autoencoder constraints to regularize visual representations, improving robustness against distribution shifts while maintaining performance and enabling interpretability.
This paper shows that mixing post-training data into pretraining (early exposure) improves how robustly a model retains capabilities after subsequent fine-tuning, challenging the notion that immediate post-training performance predicts retention. Controlled experiments on 135M and 1B models demonstrate that early exposure consistently improves the trade-off between upstream retention and downstream performance.