2012 papers
This paper challenges the standard view of superposition in neural networks by demonstrating that, unlike in idealized uncorrelated settings where interference is merely noise, realistic feature correlations allow models to arrange features so that interference becomes constructive, thereby naturally forming the semantic clusters and cyclical structures observed in real language models.
This paper presents a study of a large language model-powered "sighted guide" for blind and low vision users in social virtual reality, revealing that participants adapt their interaction from a tool-based approach when alone to a companionable relationship in the presence of others, thereby offering key design recommendations for future accessible VR guides.
This paper extends the classical "bee equation" for swarm decision-making into an agent-based model where emotional valence and arousal modulate interaction rates, revealing how emotional contagion and non-linear amplification mechanisms bias consensus outcomes and accelerate convergence in collective choices.
This paper introduces BEACON, a language-conditioned navigation system that overcomes the limitations of existing 2D image-space methods by predicting an occlusion-aware Bird's-Eye View affordance heatmap from surround-view RGB-D observations, thereby significantly improving the accuracy of inferring traversable targets in occluded regions.
This paper introduces "neural debuggers," a new class of language models that emulate traditional debugging interactions like setting breakpoints and stepping through code to enable both forward and inverse execution prediction, thereby laying the foundation for more powerful agentic coding systems and automated debugging.
This paper introduces the Overfitting-Underfitting Indicator (OUI) as an efficient, early-stage metric based on hidden neuron activation patterns to distinguish optimal learning rates in PPO actor-critic training, demonstrating its superior ability to prune unpromising runs compared to traditional criteria by revealing distinct structural signatures in actor and critic networks.
The paper proposes k-MTR, a novel framework that bypasses the traditional image reconstruction step by directly learning multi-task cardiac diagnostic features from undersampled k-space data through a shared semantic manifold, thereby eliminating reconstruction artifacts and achieving competitive performance across regression, classification, and segmentation tasks.
This paper presents a data-driven methodology using geospatial analytics and machine learning to accurately estimate and identify the drivers of spectrum demand variations across urban regions, achieving 70% predictive accuracy to support flexible spectrum access policies for future 6G networks.
The paper introduces ACADiff, an adaptive clinical-aware latent diffusion framework that synthesizes missing multimodal brain imaging data (sMRI, FDG-PET, and AV45-PET) by integrating imaging observations with GPT-4o-encoded clinical metadata, achieving superior generation quality and robust diagnostic performance even when up to 80% of modalities are missing.
This paper presents an AI-driven, data-driven framework that utilizes validated proxies from site license and crowdsourced data to accurately forecast spectrum demand across five major Canadian cities, thereby enabling more efficient dynamic spectrum planning and resource allocation for regulators and network operators.
The paper proposes MSSR, a memory-aware adaptive replay framework that estimates sample-level memory strength to dynamically schedule rehearsal intervals, effectively mitigating catastrophic forgetting while maintaining fast adaptation in continual LLM fine-tuning.
This paper introduces the Dynamics-Aware Policy Learning (DAPL) framework, which leverages explicit world modeling to learn contact-induced dynamics, enabling robots to achieve robust extrinsic dexterity in cluttered environments without hand-crafted heuristics and significantly outperforming existing manipulation methods in both simulation and real-world deployments.
This paper introduces HR-GAT, a hierarchical resolution graph attention network that leverages geospatial data to predict spectrum demand with 21% higher accuracy than baseline models, effectively addressing spatial autocorrelation challenges to enable more efficient spectrum sharing and policy-making.
This paper introduces SCENEBench, a comprehensive benchmark suite designed to evaluate Large Audio Language Models on critical non-speech and cross-lingual audio understanding tasks relevant to assistive and industrial applications, revealing significant performance gaps and latency challenges in current state-of-the-art models.
This paper introduces MA-EgoQA, a novel benchmark and dataset featuring 1,700 questions across five categories designed to evaluate the ability of AI models to understand and reason over multiple long-horizon egocentric videos from embodied agents, alongside a proposed baseline model named EgoMAS that highlights current limitations in system-level multi-agent understanding.
This paper introduces the smoothing pseudo-projector, a lightweight, multigrid-inspired module that corrects hidden representations in transformer-based models to suppress noise from label-irrelevant inputs, thereby improving training dynamics and robustness without altering the core architecture.
The paper introduces MITRA, an on-premise Retrieval-Augmented Generation (RAG) system designed to enhance knowledge retrieval in large-scale physics collaborations like CMS by employing a novel automated pipeline for document processing and a two-tiered vector database architecture to accurately answer context-aware questions while ensuring data privacy.
This paper proposes ACP-SL, an adaptive channel pruning scheme for Split Learning that utilizes a label-aware channel importance scoring module to compress smashed data, thereby significantly reducing communication overhead while improving test accuracy and training efficiency.
The paper proposes "Ego," an efficient personalization method for vision-language models that extracts visual tokens representing target concepts via internal attention mechanisms to serve as memory, enabling strong performance across single-concept, multi-concept, and video personalization tasks without requiring additional training stages or external modules.
This paper introduces EXPLORE-Bench, a benchmark derived from real first-person videos to evaluate the ability of multimodal large language models to perform long-horizon egocentric scene prediction, revealing significant performance gaps compared to humans and demonstrating that stepwise reasoning offers partial improvements at a computational cost.