arXiv subjects

Sijia Zhong

Publications and source records attributed to Sijia Zhong.

5 recordsLinked to original sources

arXiv2026

A Survey on Human-AI Collaboration with Large Foundation Models

As the capabilities of artificial intelligence (AI) continue to expand rapidly, Human-AI (HAI) Collaboration, combining human intellect and AI systems, has become pivotal for advancing problem-solving and decision-making processes. The advent of Large Foundation Models (LFMs) has greatly expanded its potential, offering unprecedented capabilities by leveraging vast amounts of data to understand and predict complex patterns. At the same time, realizing this potential responsibly requires addressing persistent challenges related to safety, fairness, and control. This paper reviews the crucial integration of LFMs with HAI, highlighting both opportunities and risks. We structure our analysis around four areas: human-guided model development, collaborative design principles, ethical and governance frameworks, and applications in high-stakes domains. Our review shows that successful HAI systems are not the automatic result of stronger models but the product of careful, human-centered design. By identifying key open challenges, this survey aims to give insight into current and future research that turns the raw power of LFMs into partnerships that are reliable, trustworthy, and beneficial to society.

cs.AI↗

arXiv2024

Right this way: Can VLMs Guide Us to See More to Answer Questions?

In question-answering scenarios, humans can assess whether the available information is sufficient and seek additional information if necessary, rather than providing a forced answer. In contrast, Vision Language Models (VLMs) typically generate direct, one-shot responses without evaluating the sufficiency of the information. To investigate this gap, we identify a critical and challenging task in the Visual Question Answering (VQA) scenario: can VLMs indicate how to adjust an image when the visual information is insufficient to answer a question? This capability is especially valuable for assisting visually impaired individuals who often need guidance to capture images correctly. To evaluate this capability of current VLMs, we introduce a human-labeled dataset as a benchmark for this task. Additionally, we present an automated framework that generates synthetic training data by simulating ``where to know'' scenarios. Our empirical results show significant performance improvements in mainstream VLMs when fine-tuned with this synthetic data. This study demonstrates the potential to narrow the gap between information assessment and acquisition in VLMs, bringing their performance closer to humans.

cs.CV↗

arXiv2014

Waiting time distribution of solar energetic particle events modeled with a non-stationary Poisson process

We present a study of the waiting time distributions (WTDs) of solar energetic particle (SEP) events observed with the spacecraft $WIND$ and $GOES$. Both the WTDs of solar electron events (SEEs) and solar proton events (SPEs) display a power-law tail $\sim Δt^{-γ}$. The SEEs display a broken power-law WTD. The power-law index is $γ_{1} =$ 0.99 for the short waiting times ($<$70 hours) and $γ_{2} =$ 1.92 for large waiting times ($>$100 hours). The break of the WTD of SEEs is probably due to the modulation of the corotating interaction regions (CIRs). The power-law index $γ\sim$ 1.82 is derived for the WTD of SPEs that is consistent with the WTD of type II radio bursts, indicating a close relationship between the shock wave and the production of energetic protons. The WTDs of SEP events can be modeled with a non-stationary Poisson process which was proposed to understand the waiting time statistics of solar flares (Wheatland 2000; Aschwanden $\&$ McTiernan 2010). We generalize the method and find that, if the SEP event rate $λ= 1/Δt$ varies as the time distribution of event rate $f(λ) = A λ^{-α}exp(-βλ)$, the time-dependent Poisson distribution can produce a power-law tail WTD $\sim Δt^{α- 3}$, where $0 \leq α< 2$.

astro-ph.SR↗

arXiv2009

Low regularity global well-posedness for the two-dimensional Zakharov system

The two-dimensional Zakharov system is shown to have a unique global solution for data without finite energy if the L^2 - norm of the Schrödinger part is small enough. The proof uses a refined I-method originally initiated by Colliander, Keel, Staffilani, Takaoka and Tao. A polynomial growth bound for the solution is also given.

math.AP↗

arXiv2007

L^2-concentration phenomenon for Zakharov system below energy norm

In this paper, we'll prove a L^2-concentration result of Zakharov system in space dimension two, with radial initial data (u_0,n_0,n_1)\in H^s\times L^2\times H^{-1} ({16/17}<s<1), when blow up of the solution happens by I-method. In additional to that we find a blow up character of this system. Furthermore, we improve the global existence result of Bourgain's to above spaces.

math.AP↗