Category: Computer Science & AI
Quantum-Enhanced In-Context Learning for Geopotential Field Estimation: A Theoretical Framework
Theoretical framework proves transformers efficiently learn Earth’s gravitational field with quantum gravimetry advantages, but inverse problems remain intractable.
A Wizard for Kids: A Platform for Improvised Child–Robot Interactions
Platform design for teleoperated social robots in classrooms, enabling safe child-robot interactions and iterative prototyping of educational robot applications.
When Robots Should Break the Rules
Framework proposing that robots deliberately breaking conventional behavioral rules can create more ethical, effective, and socially intelligent interactions than strict rule adherence.
SLAWS: Spatial Locality Analysis and Workload Orchestration for Sparse Matrix Multiplication
SLAWS framework enhances sparse matrix multiplication by analyzing data locality patterns and orchestrating workloads adaptively, overcoming limitations of fixed-architecture accelerators.
Deskly: A Privacy-First Desktop Digital Wellbeing System for Windows Using Behavioral Nudges and Gamification
Deskly, a Windows desktop wellbeing system using behavioral nudges and local data storage, reduced screen time by 23% and improved wellbeing scores by 17% in a 12-participant pilot.
SCOPE: Real-Time Natural Language Camera Agent at the Edge
SCOPE integrates natural language processing with camera control for edge deployment, executing perception and planning locally without cloud dependencies.
PAT: Accelerating LLM Decoding via P refix- A ware A t tention with Resource Efficient Multi-Tile Kernel
PAT optimizes LLM decode-phase attention by exploiting shared request prefixes and adaptive kernel tiling, reducing memory bandwidth bottlenecks in multi-request serving scenarios.
Data augmented hybrid GCN transformer for student engagement recognition in E-learning
Hybrid framework combining graph networks and transformers with synthetic data augmentation for automatic student engagement recognition from facial video in e-learning systems.
Human-centric Evaluation of Semantic Resources: A Systematic Mapping Study
Systematic mapping of human-centric evaluation approaches for semantic resources like ontologies and knowledge graphs, synthesizing 15 years of research into a theoretical framework.
CORE: Data Augmentation for Link Prediction via Information Bottleneck
CORE applies Information Bottleneck principles to augment graph data for link prediction, simultaneously recovering missing edges and reducing noise to enhance model robustness.
