arXiv subjects

Zhongjian Hu

Publications and source records attributed to Zhongjian Hu.

4 recordsLinked to original sources

arXiv2025

Rationale-guided Prompting for Knowledge-based Visual Question Answering

Recently, Large Language Models (LLMs) have been used for knowledge-based Visual Question Answering (VQA). Despite the encouraging results of previous studies, prior methods prompt LLMs to predict answers directly, neglecting intermediate thought processes. We argue that prior methods do not sufficiently activate the capacities of LLMs. We propose a framework called PLRH that Prompts LLMs with Rationale Heuristics for knowledge-based VQA. The PLRH prompts LLMs with Chain of Thought (CoT) to generate rationale heuristics, i.e., intermediate thought processes, and then leverages the rationale heuristics to inspire LLMs to predict answers. Experiments show that our approach outperforms the existing baselines by more than 2.2 and 2.1 on OK-VQA and A-OKVQA, respectively.

cs.CL↗

arXiv2025

Multi-Agents Based on Large Language Models for Knowledge-based Visual Question Answering

Large Language Models (LLMs) have achieved impressive results in knowledge-based Visual Question Answering (VQA). However existing methods still have challenges: the inability to use external tools autonomously, and the inability to work in teams. Humans tend to know whether they need to use external tools when they encounter a new question, e.g., they tend to be able to give a direct answer to a familiar question, whereas they tend to use tools such as search engines when they encounter an unfamiliar question. In addition, humans also tend to collaborate and discuss with others to get better answers. Inspired by this, we propose the multi-agent voting framework. We design three LLM-based agents that simulate different levels of staff in a team, and assign the available tools according to the levels. Each agent provides the corresponding answer, and finally all the answers provided by the agents are voted to get the final answer. Experiments on OK-VQA and A-OKVQA show that our approach outperforms other baselines by 2.2 and 1.0, respectively.

cs.CL↗

arXiv2021

Single Photon Sources with Near Unity Collection Efficiencies by Deterministic Placement of Quantum Dots in Nanoantennas

Deterministic coupling between photonic nodes in a quantum network is an essential step towards implementing various quantum technologies. The omnidirectionality of free-standing emitters, however, makes this coupling highly inefficient, in particular if the distant nodes are coupled via low numerical aperture (NA) channels such as optical fibers. This limitation requires placing quantum emitters in nanoantennas that can direct the photons into the channels with very high efficiency. Moreover, to be able to scale such technologies to a large number of channels, the placing of the emitters should be deterministic. In this work we present a method for directly locating single free-standing quantum emitters with high spatial accuracy at the center of highly directional bullseye metal-dielectric nanoantennas. We further employ non-blinking, high quantum yield colloidal quantum dots (QDs) for on-demand single-photon emission that is uncompromised by instabilities or non-radiative exciton recombination processes. Taken together this approach results in a record-high collection efficiency of 85% of the single photons into a low NA of 0.5, setting the stage for efficient coupling between on-chip, room temperature nanoantenna-emitter devices and a fiber or a remote free-space node without the need for additional optics.

quant-ph↗

arXiv2018

Influence of Morphology on Blinking Mechanisms and Excitonic Fine Structure of Single Colloidal Nanoplatelets

Colloidal semiconductor nanoplatelets (NPLs) with electronic structure as quantum wells have recently emerged as exciting materials for optoelectronic applications. Here we investigate how morphology affects important photoluminescence (PL) properties of single CdSe and core/shell CdSe/CdZnS nanoplatelets. By analyzing PL intensity-lifetime correlation and second-order photon correlation results, we demonstrate that, irrespective of morphology, Auger recombination cannot be responsible for PL blinking of single NPLs. We propose that hot carrier trapping plays a significant role in blinking and find that a rough shell induces additional nonradiative channels presumably related to defects or traps of an imperfect shell. Polarization-resolved PL spectroscopy analysis reveals exciton fine-structure splitting on the order of several tens of meV in rough-shell NPLs at room temperature, which is attributed to exciton localization and substantiated with theoretical calculations taking into account the NPL shape and electron-hole exchange interaction.

cond-mat.mes-hall↗