Archive, newest first. Filter by topic, impact and company →
Problem This work addresses the gap in predictive modeling for water-surface elevation (WSE) in HEC-RAS 2D simulations, specifically focusing on the limitations of existing surrogate models that often rely on…
Problem The paper addresses the gap in understanding whether artificial systems can exhibit consciousness, highlighting limitations in existing methodologies that either rely on discriminative checklists or directly engineer consciousness-inspired modules.…
Problem This paper addresses the limited sensitivity of medical vision-language models (VLMs) to subtle lesions in clinical images, which are often characterized by sparse, low-contrast visual evidence embedded within complex…
Problem The paper addresses the challenge of limited annotated data in the digital twin (DT) framework for infrastructure inspection, particularly in the context of traffic sign monitoring. The authors highlight…
Problem The paper addresses the limitations of existing wireless foundation models that utilize masked input reconstruction, which tends to bias learned representations towards low-level signal details. This work is particularly…
Problem The paper addresses the limitations of existing Scene Graph Generation (SGG) models, which perform well on frequent relation types but struggle with annotation sparsity, leading to unreliable predictions. The…
Problem The paper addresses the challenge of subgraph detection, which is crucial across various scientific domains but is hindered by the NP-completeness of subgraph isomorphism. Traditional combinatorial approaches are limited…
Problem Urban green-space extraction from ultra-high-resolution (UHR) imagery typically relies on patch-wise analysis, which restricts the semantic reuse of visually similar vegetation patterns across spatially separated areas. Existing methods that…
Problem The paper addresses the gap in physical consistency in image-to-video diffusion models, which often generate motion that violates physical laws. Despite advancements in visual fidelity, existing models struggle with…
Problem This work addresses the gap in modeling individual decision-making during infectious disease outbreaks, which is crucial for understanding behavioral dynamics and informing public health interventions. Prior research has utilized…
Problem — This study addresses the gap in effective methodologies for mapping bacterial leaf blight (BLB) in rice using UAV multispectral imagery. Despite the growing interest in remote sensing for…
Problem The paper addresses the challenge of integrating understanding and generation capabilities in audio processing, specifically focusing on the limitations of continuous audio autoencoders, which produce latents with insufficient structure…
Problem The paper addresses the gap in existing multimodal generative models that struggle to reliably incorporate structured, domain-specific, or safety-critical knowledge during generation. Current methods, such as prompt augmentation and…
Problem This work addresses the limitations of existing machine learning approaches in football tactics, which primarily focus on analyzing historical actions or predefined counterfactual scenarios. The authors highlight a gap…
Problem The paper addresses the challenge of vessel trajectory prediction from Automatic Identification System (AIS) data, which is critical for maritime situational awareness. Current methods struggle with irregular sampling, missing…
Problem The paper addresses the gap in reliable rubric grading for large language models (LLMs), emphasizing that accurate score prediction alone is insufficient. Existing methods for credit assignment and intervention…
Problem This work addresses the significant gap in Natural Language Processing (NLP) resources for Lombard, an under-resourced language continuum in Italy. The authors highlight the lack of high-quality datasets necessary…
Problem This study addresses the gap in regional performance evaluation of Machine Learning Weather Prediction (MLWP) models, specifically GraphCast, in the Global South, where existing literature is limited and unverified.…
Problem The paper addresses the critical gap in attack detection for cyber-physical systems (CPS) where the plant model or its structure is unknown. This is particularly relevant in scenarios where…
Problem The paper addresses the challenge of limited labeled neural data in brain decoding, particularly in low-data regimes. This issue is critical as it hampers the development of effective decoding…
Problem The paper addresses the limitations of existing equivariant networks in probabilistic inference over spatially embedded variables, particularly the inability to produce rank-2 precision tensors necessary for representing anisotropic uncertainty.…
Problem This work addresses the limitations of existing paired topological divergences, particularly the Representation Topology Divergence (RTD), which suffers from heuristic asymmetry and unbounded scores that vary with sample size.…
Problem Existing Video Question Answering (VideoQA) systems excel at factoid questions but struggle with deep video understanding (DVU), which necessitates comprehension of complex narratives across long video content. The authors…
Problem Estimating local mean curvature in high-dimensional datasets is crucial for geometry-aware machine learning algorithms, such as the Mean Curvature Boundary Points (MCBP) method. The naive approach, which relies on…
Problem The paper addresses the challenge of machine unlearning, specifically for autoregressive language models, where the goal is to remove specific knowledge while maintaining overall model performance. Existing methods either…
Problem This paper addresses the high inference latency and computational costs associated with video generation models based on Diffusion Transformers (DiTs), which are hindered by the quadratic complexity of 3D…
Problem Factual sycophancy, where language models provide incorrect answers under social pressure, remains inadequately understood in terms of its underlying mechanisms. This paper addresses the gap in the literature by…
Problem This work addresses the gap in the literature regarding One-to-Many Temporal Grounding (OMTG), a scenario where multiple disjoint video segments must be localized for a single textual query. Prior…
Problem — The paper addresses the challenge of robotic manipulation of textiles, particularly the difficulties posed by continuous deformation and self-occlusions that hinder robust visual perception. The authors highlight the…
Problem Existing evaluations of memorization in large language models (LLMs) primarily focus on the models' ability to reproduce training data under adversarial conditions, rather than assessing their propensity to do…
