MedVIGIL evaluates whether medical vision-language models recognize when visual evidence no longer supports an answer. The clinician-supervised benchmark tests false premises, wording changes, and image perturbations, measuring safe refusal and silent failure alongside answer correctness.

SYNAPSE organizes episodic and semantic memories in a unified graph. Spreading activation, lateral inhibition, and temporal decay identify relevant subgraphs, while hybrid retrieval combines graph activation with semantic similarity to support long-term agent memory.

ViThinker enables vision-language models to generate perceptual queries during reasoning. A two-stage training curriculum distills frozen vision experts and learns task-driven querying, allowing the model to synthesize relevant visual features without external tool calls at inference time.

This survey examines the bidirectional relationship between quantum computing and artificial intelligence, covering AI for quantum hardware and algorithms, quantum methods for learning, and the practical challenges of hybrid quantum-classical systems.

ADLGen synthesizes symbolic, event-triggered sensor sequences for human activity modeling, providing a novel approach to activity recognition and modeling.

We introduce Adaptive Quantum-Classical Fusion (AQCF), the first framework to bridge quantum and classical computing through dynamic, quantum-classical co-design for next-generation language models.

Quantum Autoencoder for Molecular Representation Learning.

Leveraging Multi-View Images for Improved 3D Scene Understanding with Large Language Models.

Eye-Gaze Guided Transformer on Spiking Neural Networks for Medical Image Analysis.

We propose EGMA, a novel framework for medical multi-modal alignment, marking the first attempt to integrate eye-gaze data into vision-language pre-training. EGMA outperforms existing state-of-the-art medical multi-modal pre-training methods, and realizes notable enhancements in image classification and image-text retrieval tasks. EGMA demonstrates that even a small amount of eye-gaze data can effectively assist in multi-modal pre-training and improve the feature representation ability of the model.