Institution profile

Brookhaven National Laboratory

Academic institutionnorthamerica · us
Official website
Research library134linked papers
Opportunities0open roles
Selected work

Representative Papers

GFlowNet Training by Policy Gradients

Aug 12, 2024International Conference on Machine Learning

GFlowNets suffer from low training efficiency and unstable gradient estimation in combinatorial object generation due to strict flow conservation constraints. To address this, we propose the first policy-gradient-based GFlowNet training framework. Our method reformulates flow conservation as a policy optimization objective, enabling joint training of forward and backward policies without explicit flow matching. We provide theoretical convergence guarantees and introduce a coupled update mechanism to reduce gradient variance. Experiments across multiple synthetic and real-world datasets demonstrate that our approach significantly improves sample quality, training stability, and fidelity to the target distribution—particularly under sparse-reward settings, where it exhibits superior robustness.

3 citationsRead paper

Quantum Super-resolution by Adaptive Non-local Observables

Jan 20, 2026

This work proposes a novel quantum approach to image super-resolution by introducing variational quantum circuits (VQCs) into the task for the first time. To overcome the limitations of conventional deep learning methods—which often require extensive data and computational resources while struggling to capture fine-grained correlations—the authors develop an Adaptive Non-local Observable (ANO) mechanism. This mechanism employs trainable multi-qubit Hermitian operators to dynamically adjust the measurement process, thereby transcending the constraints of fixed Pauli readouts. By harnessing quantum superposition and entanglement within a high-dimensional Hilbert space, the model effectively learns the mapping from low- to high-resolution images. Experimental results demonstrate that the proposed method achieves up to a 5× resolution enhancement with a significantly smaller model footprint, highlighting the promising potential of quantum machine learning for super-resolution tasks.

1 citationsRead paper

Mem-Gallery: Benchmarking Multimodal Long-Term Conversational Memory for MLLM Agents

Jan 07, 2026arXiv.org

Existing benchmarks struggle to effectively evaluate the ability of multimodal large language models to retain, organize, and evolve visual and textual information over extended dialogues. To address this gap, this work proposes Mem-Gallery—the first comprehensive benchmark specifically designed for assessing multimodal long-term dialogue memory. It features high-quality, multi-turn image-text conversations and introduces a three-dimensional evaluation framework encompassing memory retrieval and adaptation, memory-based reasoning, and knowledge management. Evaluations of 13 representative memory systems using Mem-Gallery reveal critical bottlenecks in current models’ capacity for organizing and reasoning over multimodal memories, underscoring the necessity of explicitly preserving and structurally organizing multimodal information to support robust long-term dialogue capabilities.

1 citationsRead paper
Recent publications

Latest Papers