arXiv subjects

Yiyang Jiang

Publications and source records attributed to Yiyang Jiang.

25 records · Page 2Linked to original sources

arXiv2024

Prior Knowledge Integration via LLM Encoding and Pseudo Event Regulation for Video Moment Retrieval

In this paper, we investigate the feasibility of leveraging large language models (LLMs) for integrating general knowledge and incorporating pseudo-events as priors for temporal content distribution in video moment retrieval (VMR) models. The motivation behind this study arises from the limitations of using LLMs as decoders for generating discrete textual descriptions, which hinders their direct application to continuous outputs like salience scores and inter-frame embeddings that capture inter-frame relations. To overcome these limitations, we propose utilizing LLM encoders instead of decoders. Through a feasibility study, we demonstrate that LLM encoders effectively refine inter-concept relations in multimodal embeddings, even without being trained on textual embeddings. We also show that the refinement capability of LLM encoders can be transferred to other embeddings, such as BLIP and T5, as long as these embeddings exhibit similar inter-concept similarity patterns to CLIP embeddings. We present a general framework for integrating LLM encoders into existing VMR architectures, specifically within the fusion module. Through experimental validation, we demonstrate the effectiveness of our proposed methods by achieving state-of-the-art performance in VMR. The source code can be accessed at https://github.com/fletcherjiang/LLMEPET.

cs.CV↗

arXiv2024

Hybrid-order topology in unconventional magnets of Eu-based Zintl compounds with surface-dependent quantum geometry

The exploration of magnetic topological insulators is instrumental in exploring axion electrodynamics and intriguing transport phenomena, such as the quantum anomalous Hall effect. Here, we report that a family of magnetic compounds Eu$_{2n+1}$In$_{2}$(As,Sb)$_{2n+2}$ ($n=0,1,2$) exhibit both gapless Dirac surface states and chiral hinge modes. Such a hybrid-order topology hatches surface-dependent quantum geometry. By mapping the responses into real space, we demonstrate the existence of chiral hinge modes along the $c$ direction, which originate from the half-quantized anomalous Hall effect on two gapped $ac$/$bc$ facets due to Berry curvature, while the unpinned Dirac surface states on the gapless $ab$ facet generate an intrinsic nonlinear anomalous Hall effect due to the quantum metric. When Eu$_{3}$In$_{2}$As$_{4}$ is polarized to the ferromagnetic phase by an external magnetic field, it becomes an ideal Weyl semimetal with a single pair of type-I Weyl points and no extra Fermi pocket. Our work predicts rich topological states sensitive to magnetic structures, quantum geometry-induced transport and topological superconductivity if proximitized with a superconductor.

cond-mat.mtrl-sci↗

arXiv2024

Three-dimensional higher-order saddle points induced flat bands in Co-based kagome metals

The saddle point (van Hove singularity) exhibits a divergent density of states in 2D systems, leading to fascinating phenomena like strong correlations and unconventional superconductivity, yet it is seldom observed in 3D systems. In this work, we have found two types of 3D higher-order saddle points (HOSPs) in emerging 3D kagome metals, YbCo$_6$Ge$_6$ and MgCo$_6$Ge$_6$. Both HOSPs exhibit a singularity in their density of states, which is significantly enhanced compared to the ordinary saddle point. The HOSP near the Fermi energy generates a flat band extending a large area in the Brillouin zone, potentially amplifying the correlation effect and fostering electronic instabilities. Two types of HOSPs exhibit distinct robustness upon element substitution and lattice distortions in these kagome compounds. Our work paves the way for engineering exotic band structures, such as saddle points and flat bands, and exploring interesting phenomena in Co-based kagome materials.

cond-mat.mtrl-sci↗

arXiv2024

LibriSQA: A Novel Dataset and Framework for Spoken Question Answering with Large Language Models

While Large Language Models (LLMs) have demonstrated commendable performance across a myriad of domains and tasks, existing LLMs still exhibit a palpable deficit in handling multimodal functionalities, especially for the Spoken Question Answering (SQA) task which necessitates precise alignment and deep interaction between speech and text features. To address the SQA challenge on LLMs, we initially curated the free-form and open-ended LibriSQA dataset from Librispeech, comprising Part I with natural conversational formats and Part II encompassing multiple-choice questions followed by answers and analytical segments. Both parts collectively include 107k SQA pairs that cover various topics. Given the evident paucity of existing speech-text LLMs, we propose a lightweight, end-to-end framework to execute the SQA task on the LibriSQA, witnessing significant results. By reforming ASR into the SQA format, we further substantiate our framework's capability in handling ASR tasks. Our empirical findings bolster the LLMs' aptitude for aligning and comprehending multimodal information, paving the way for the development of universal multimodal LLMs. The dataset and demo can be found at https://github.com/ZihanZhaoSJTU/LibriSQA.

cs.CL↗

arXiv2024

Chirality-Induced Magnet-Free Spin Generation in a Semiconductor

Electrical generation and transduction of polarized electron spins in semiconductors are of central interest in spintronics and quantum information science. While spin generation in semiconductors has been frequently realized via electrical injection from a ferromagnet, there are significant advantages in nonmagnetic pathways of creating spin polarization. One such pathway exploits the interplay of electron spin with chirality in electronic structures or real space. Here, utilizing chirality-induced spin selectivity (CISS), we demonstrate efficient creation of spin accumulation in n-doped GaAs via electric current injection from a normal metal (Au) electrode through a self-assembled monolayer of chiral molecules (α-helix L-polyalanine, AHPA-L). The resulting spin polarization is detected as a Hanle effect in the n-GaAs, which is found to obey a distinct universal scaling with temperature and bias current consistent with chirality-induced spin accumulation. The experiment constitutes a definitive observation of CISS in a fully nonmagnetic device structure and demonstration of its ability to generate spin accumulation in a conventional semiconductor. The results thus place key constraints on the physical mechanism of CISS and present a new scheme for magnet-free semiconductor spintronics.

cond-mat.mes-hall↗

arXiv2023

Tail probability of maximal displacement in critical branching Lévy process with stable branching

Consider a critical branching Lévy process $\{X_t, t\ge 0\}$ with branching rate $β>0, $ offspring distribution $\{p_k:k\geq 0\}$ and spatial motion $\{ξ_t, Π_x\}$. For any $t\ge 0$, let $N_t$ be the collection of particles alive at time $t$, and, for any $u\in N_t$, let $X_u(t)$ be the position of $u$ at time $t$. We study the tail probability of the maximal displacement $M:=\sup_{t>0}\sup_{u\in N_t} X_u(t)$ under the assumption $\lim_{n\to\infty} n^α\sum_{k=n}^\infty p_k =κ\in(0,\infty)$ for some $α\in (1,2)$, $Π_0(ξ_1)=0$ and $Π_0 (|ξ_1|^r)\in (0,\infty)$ for some $r> 2α/(α-1)$. Our main result is a generalization of the main result of Sawyer and Fleischman (1979) for branching Brownian motions and that of Lalley and Shao (2015) for branching random walks, both of which are proved under the assumption $\sum_{k=0}^\infty k^3 p_k<\infty$.

math.PR↗

arXiv2022

Shotgun threshold for sparse Erdős-Rényi graphs

In the shotgun assembly problem for a graph, we are given the empirical profile for rooted neighborhoods of depth $r$ (up to isomorphism) for some $r\geq 1$ and we wish to recover the underlying graph up to isomorphism. When the underlying graph is an Erdős-Rényi $\mathcal G(n, \fracλ{n})$, we show that the shotgun assembly threshold $r_* \approx \frac{ \log n}{\log (λ^2 γ_λ)^{-1}}$ where $γ_λ$ is the probability for two independent Poisson-Galton-Watson trees with parameter $λ$ to be rooted isomorphic with each other. Our result sharpens a constant factor in a previous work by Mossel and Ross (2019) and thus solves a question therein.

math.PR↗