arXiv subjects

Yong Li

Publications and source records attributed to Yong Li.

At least 505 records · Page 28Linked to original sources

arXiv2024

From Pixels to Progress: Generating Road Network from Satellite Imagery for Socioeconomic Insights in Impoverished Areas

The Sustainable Development Goals (SDGs) aim to resolve societal challenges, such as eradicating poverty and improving the lives of vulnerable populations in impoverished areas. Those areas rely on road infrastructure construction to promote accessibility and economic development. Although publicly available data like OpenStreetMap is available to monitor road status, data completeness in impoverished areas is limited. Meanwhile, the development of deep learning techniques and satellite imagery shows excellent potential for earth monitoring. To tackle the challenge of road network assessment in impoverished areas, we develop a systematic road extraction framework combining an encoder-decoder architecture and morphological operations on satellite imagery, offering an integrated workflow for interdisciplinary researchers. Extensive experiments of road network extraction on real-world data in impoverished regions achieve a 42.7% enhancement in the F1-score over the baseline methods and reconstruct about 80% of the actual roads. We also propose a comprehensive road network dataset covering approximately 794,178 km2 area and 17.048 million people in 382 impoverished counties in China. The generated dataset is further utilized to conduct socioeconomic analysis in impoverished counties, showing that road network construction positively impacts regional economic development. The technical appendix, code, and generated dataset can be found at https://github.com/tsinghua-fib-lab/Road_network_extraction_impoverished_counties.

cs.CV↗

arXiv2024

Deep Learning for Cross-Domain Data Fusion in Urban Computing: Taxonomy, Advances, and Outlook

As cities continue to burgeon, Urban Computing emerges as a pivotal discipline for sustainable development by harnessing the power of cross-domain data fusion from diverse sources (e.g., geographical, traffic, social media, and environmental data) and modalities (e.g., spatio-temporal, visual, and textual modalities). Recently, we are witnessing a rising trend that utilizes various deep-learning methods to facilitate cross-domain data fusion in smart cities. To this end, we propose the first survey that systematically reviews the latest advancements in deep learning-based data fusion methods tailored for urban computing. Specifically, we first delve into data perspective to comprehend the role of each modality and data source. Secondly, we classify the methodology into four primary categories: feature-based, alignment-based, contrast-based, and generation-based fusion methods. Thirdly, we further categorize multi-modal urban applications into seven types: urban planning, transportation, economy, public safety, society, environment, and energy. Compared with previous surveys, we focus more on the synergy of deep learning methods with urban computing applications. Furthermore, we shed light on the interplay between Large Language Models (LLMs) and urban computing, postulating future research directions that could revolutionize the field. We firmly believe that the taxonomy, progress, and prospects delineated in our survey stand poised to significantly enrich the research community. The summary of the comprehensive and up-to-date paper list can be found at https://github.com/yoshall/Awesome-Multimodal-Urban-Computing.

cs.LG↗

arXiv2024

Effectively Compress KV Heads for LLM

The advent of pre-trained large language models (LLMs) has revolutionized various natural language processing tasks. These models predominantly employ an auto-regressive decoding mechanism that utilizes Key-Value (KV) caches to eliminate redundant calculations for previous tokens. Nevertheless, as context lengths and batch sizes increase, the linear expansion in memory footprint of KV caches becomes a key bottleneck of LLM deployment, which decreases generation speeds significantly. To mitigate this issue, previous techniques like multi-query attention (MQA) and grouped-query attention (GQA) have been developed, in order to reduce KV heads to accelerate inference with comparable accuracy to multi-head attention (MHA). Despite their effectiveness, existing strategies for compressing MHA often overlook the intrinsic properties of the KV caches. In this work, we explore the low-rank characteristics of the KV caches and propose a novel approach for compressing KV heads. In particular, we carefully optimize the MHA-to-GQA transformation to minimize compression error, and to remain compatible with rotary position embeddings (RoPE), we also introduce specialized strategies for key caches with RoPE. We demonstrate that our method can compress half or even three-quarters of KV heads while maintaining performance comparable to the original LLMs, which presents a promising direction for more efficient LLM deployment in resource-constrained environments.

cs.CL↗

arXiv2024

Partial train and isolate, mitigate backdoor attack

Neural networks are widely known to be vulnerable to backdoor attacks, a method that poisons a portion of the training data to make the target model perform well on normal data sets, while outputting attacker-specified or random categories on the poisoned samples. Backdoor attacks are full of threats. Poisoned samples are becoming more and more similar to corresponding normal samples, and even the human eye cannot easily distinguish them. On the other hand, the accuracy of models carrying backdoors on normal samples is no different from that of clean models.In this article, by observing the characteristics of backdoor attacks, We provide a new model training method (PT) that freezes part of the model to train a model that can isolate suspicious samples. Then, on this basis, a clean model is fine-tuned to resist backdoor attacks.

cs.CV↗

arXiv2024

Chain-of-Planned-Behaviour Workflow Elicits Few-Shot Mobility Generation in LLMs

The powerful reasoning capabilities of large language models (LLMs) have brought revolutionary changes to many fields, but their performance in human behaviour generation has not yet been extensively explored. This gap likely emerges because the internal processes governing behavioral intentions cannot be solely explained by abstract reasoning. Instead, they are also influenced by a multitude of factors, including social norms and personal preference. Inspired by the Theory of Planned Behaviour (TPB), we develop a LLM workflow named Chain-of-Planned Behaviour (CoPB) for mobility behaviour generation, which reflects the important spatio-temporal dynamics of human activities. Through exploiting the cognitive structures of attitude, subjective norms, and perceived behaviour control in TPB, CoPB significantly enhance the ability of LLMs to reason the intention of next movement. Specifically, CoPB substantially reduces the error rate of mobility intention generation from 57.8% to 19.4%. To improve the scalability of the proposed CoPB workflow, we further explore the synergy between LLMs and mechanistic models. We find mechanistic mobility models, such as gravity model, can effectively map mobility intentions to physical mobility behaviours. The strategy of integrating CoPB with gravity model can reduce the token cost by 97.7% and achieve better performance simultaneously. Besides, the proposed CoPB workflow can facilitate GPT-4-turbo to automatically generate high quality labels for mobility behavior reasoning. We show such labels can be leveraged to fine-tune the smaller-scale, open source LLaMA 3-8B, which significantly reduces usage costs without sacrificing the quality of the generated behaviours.

cs.AI↗

arXiv2024

Llumnix: Dynamic Scheduling for Large Language Model Serving

Inference serving for large language models (LLMs) is the key to unleashing their potential in people's daily lives. However, efficient LLM serving remains challenging today because the requests are inherently heterogeneous and unpredictable in terms of resource and latency requirements, as a result of the diverse applications and the dynamic execution nature of LLMs. Existing systems are fundamentally limited in handling these characteristics and cause problems such as severe queuing delays, poor tail latencies, and SLO violations. We introduce Llumnix, an LLM serving system that reacts to such heterogeneous and unpredictable requests by runtime rescheduling across multiple model instances. Similar to context switching across CPU cores in modern operating systems, Llumnix reschedules requests to improve load balancing and isolation, mitigate resource fragmentation, and differentiate request priorities and SLOs. Llumnix implements the rescheduling with an efficient and scalable live migration mechanism for requests and their in-memory states, and exploits it in a dynamic scheduling policy that unifies the multiple rescheduling scenarios elegantly. Our evaluations show that Llumnix improves tail latencies by an order of magnitude, accelerates high-priority requests by up to 1.5x, and delivers up to 36% cost savings while achieving similar tail latencies, compared against state-of-the-art LLM serving systems. Llumnix is publicly available at https://github.com/AlibabaPAI/llumnix.

cs.AR↗

arXiv2024

Existence, Uniqueness and Asymptotic Dynamics of Nonlinear Schrödinger Equations With Quasi-Periodic Initial Data: II. The Derivative NLS

This is the second part of a two-paper series studying the nonlinear Schrödinger equation with quasi-periodic initial data. In this paper, we focus on the quasi-periodic Cauchy problem for the derivative nonlinear Schrödinger equation. Under the assumption that the Fourier coefficients of the initial data obey an exponential upper bound, we establish local existence of a solution that retains quasi-periodicity in space with a slightly weaker Fourier decay. Moreover, the solution is shown to be unique within this class of quasi-periodic functions. Also, we prove that, for the derivative nonlinear Schrödinger equation in a weakly nonlinear setting, within the time scale, as the small parameter of nonlinearity tends to zero, the nonlinear solution converges asymptotically to the linear solution in the sense of both sup-norm and analytic Sobolev-norm. The proof proceeds via a consideration of an associated infinite system of coupled ordinary differential equations for the Fourier coefficients and an explicit combinatorial analysis for the Picard iteration with the help of Feynman diagrams and the power of $\ast^{[\cdot]}$ labelling the complex conjugate.

math.AP↗

arXiv2024

DFAMiner: Mining minimal separating DFAs from labelled samples

We propose DFAMiner, a passive learning tool for learning minimal separating deterministic finite automata (DFA) from a set of labelled samples. Separating automata are an interesting class of automata that occurs generally in regular model checking and has raised interest in foundational questions of parity game solving. We first propose a simple and linear-time algorithm that incrementally constructs a three-valued DFA (3DFA) from a set of labelled samples given in the usual lexicographical order. This 3DFA has accepting and rejecting states as well as don't-care states, so that it can exactly recognise the labelled examples. We then apply our tool to mining a minimal separating DFA for the labelled samples by minimising the constructed automata via a reduction to solving SAT problems. Empirical evaluation shows that our tool outperforms current state-of-the-art tools significantly on standard benchmarks for learning minimal separating DFAs from samples. Progress in the efficient construction of separating DFAs can also lead to finding the lower bound of parity game solving, where we show that DFAMiner can create optimal separating automata for simple languages with up to 7 colours. Future improvements might offer inroads to better data structures.

cs.FL↗

arXiv2024

Masked Face Recognition with Generative-to-Discriminative Representations

Masked face recognition is important for social good but challenged by diverse occlusions that cause insufficient or inaccurate representations. In this work, we propose a unified deep network to learn generative-to-discriminative representations for facilitating masked face recognition. To this end, we split the network into three modules and learn them on synthetic masked faces in a greedy module-wise pretraining manner. First, we leverage a generative encoder pretrained for face inpainting and finetune it to represent masked faces into category-aware descriptors. Attribute to the generative encoder's ability in recovering context information, the resulting descriptors can provide occlusion-robust representations for masked faces, mitigating the effect of diverse masks. Then, we incorporate a multi-layer convolutional network as a discriminative reformer and learn it to convert the category-aware descriptors into identity-aware vectors, where the learning is effectively supervised by distilling relation knowledge from off-the-shelf face recognition model. In this way, the discriminative reformer together with the generative encoder serves as the pretrained backbone, providing general and discriminative representations towards masked faces. Finally, we cascade one fully-connected layer following by one softmax layer into a feature classifier and finetune it to identify the reformed identity-aware vectors. Extensive experiments on synthetic and realistic datasets demonstrate the effectiveness of our approach in recognizing masked faces.

cs.CV↗

arXiv2024

EconAgent: Large Language Model-Empowered Agents for Simulating Macroeconomic Activities

The advent of artificial intelligence has led to a growing emphasis on data-driven modeling in macroeconomics, with agent-based modeling (ABM) emerging as a prominent bottom-up simulation paradigm. In ABM, agents (e.g., households, firms) interact within a macroeconomic environment, collectively generating market dynamics. Existing agent modeling typically employs predetermined rules or learning-based neural networks for decision-making. However, customizing each agent presents significant challenges, complicating the modeling of agent heterogeneity. Additionally, the influence of multi-period market dynamics and multifaceted macroeconomic factors are often overlooked in decision-making processes. In this work, we introduce EconAgent, a large language model-empowered agent with human-like characteristics for macroeconomic simulation. We first construct a simulation environment that incorporates various market dynamics driven by agents' decisions regarding work and consumption. Through the perception module, we create heterogeneous agents with distinct decision-making mechanisms. Furthermore, we model the impact of macroeconomic trends using a memory module, which allows agents to reflect on past individual experiences and market dynamics. Simulation experiments show that EconAgent can make realistic decisions, leading to more reasonable macroeconomic phenomena compared to existing rule-based or learning-based agents. Our codes are released at https://github.com/tsinghua-fib-lab/ACL24-EconAgent.

cs.AI↗

arXiv2024

Modeling User Fatigue for Sequential Recommendation

Recommender systems filter out information that meets user interests. However, users may be tired of the recommendations that are too similar to the content they have been exposed to in a short historical period, which is the so-called user fatigue. Despite the significance for a better user experience, user fatigue is seldom explored by existing recommenders. In fact, there are three main challenges to be addressed for modeling user fatigue, including what features support it, how it influences user interests, and how its explicit signals are obtained. In this paper, we propose to model user Fatigue in interest learning for sequential Recommendations (FRec). To address the first challenge, based on a multi-interest framework, we connect the target item with historical items and construct an interest-aware similarity matrix as features to support fatigue modeling. Regarding the second challenge, built upon feature cross, we propose a fatigue-enhanced multi-interest fusion to capture long-term interest. In addition, we develop a fatigue-gated recurrent unit for short-term interest learning, with temporal fatigue representations as important inputs for constructing update and reset gates. For the last challenge, we propose a novel sequence augmentation to obtain explicit fatigue signals for contrastive learning. We conduct extensive experiments on real-world datasets, including two public datasets and one large-scale industrial dataset. Experimental results show that FRec can improve AUC and GAUC up to 0.026 and 0.019 compared with state-of-the-art models, respectively. Moreover, large-scale online experiments demonstrate the effectiveness of FRec for fatigue reduction. Our codes are released at https://github.com/tsinghua-fib-lab/SIGIR24-FRec.

cs.IR↗

arXiv2024

MOSS: A Large-scale Open Microscopic Traffic Simulation System

In the research of Intelligent Transportation Systems (ITS), traffic simulation is a key procedure for the evaluation of new methods and optimization of strategies. However, existing traffic simulation systems face two challenges. First, how to balance simulation scale with realism is a dilemma. Second, it is hard to simulate realistic results, which requires realistic travel demand data and simulator. These problems limit computer-aided optimization of traffic management strategies for large-scale road networks and reduce the usability of traffic simulations in areas where real-world travel demand data are lacking. To address these problems, we design and implement MObility Simulation System (MOSS). MOSS adopts GPU acceleration to significantly improve the efficiency and scale of microscopic traffic simulation, which enables realistic and fast simulations for large-scale road networks. It provides realistic travel Origin-Destination (OD) matrices generation through a pre-trained generative neural network model based on publicly available data on a global scale, such as satellite imagery, to help researchers build meaningful travel demand data. It also provides a complete open toolchain to help users with road network construction, demand generation, simulation, and result analysis. The whole toolchain including the simulator can be accessed at https://moss.fiblab.net and the codes are open-source for community collaboration.

cs.DC↗

arXiv2024

Single-photon scattering in giant-atom waveguide systems with chiral coupling

We study single-photon scattering spectra of a giant atom chirally coupled to a one-dimensional waveguide at multiple connection points, and examine chirality induced effects in the scattering spectra. We show that the transmission spectra typically possess an anti-Lorentzian lineshape with a nonzero minimum, but by engineering the chirality of the multi-point coupling, the transmission spectrum of an incident photon can undergo a transition from complete transmission to total reflection at multiple frequency ``windows'', where the width of the anti-Lorentzian lineshape for each of the window can be flexibly tuned at a fixed frequency detuning. Moreover, we show that a perfect nonreciprocal photon scattering can be achieved due to the interplay between internal atomic spontaneous emission and the chirally external decay to the waveguide, in contrast to that induced by the non-Markovian retardation effect. We also consider the non-Markovian retardation effect on the scattering spectra, which allows for a photonic band gap even with only two chiral coupling points. The giant-atom-waveguide system with chiral coupling is a promising candidate for realizing single-photon routers with multiple channels.

quant-ph↗

arXiv2024

Stochastic functional partial differential equations with monotone coefficients: Poisson stability measures, exponential mixing and limit theorems

This paper examines Poisson stable (including stationary, periodic, almost periodic, Levitan almost periodic, Bohr almost automorphic, pseudo-periodic, Birkhoff recurrent, pseudo-recurrent, etc.) measures and limit theorems for stochastic functional partial differential equations(SFPDEs) with monotone coefficients. We first show the existence and uniqueness of entrance measure $μ_{t}$ for SFPDEs by dissipative method (or remoting start). Then, with the help of Shcherbakov's comparability method in character of recurrence, we prove that the entrance measure inherits the same recurrence of coefficients. Thirdly, we show the tightness of the set of measures $μ_{t}$. As a result, any sequence of the average of $\{μ_{t}\}_{t\in\mathbb{R} }$ have the limit point $μ^{*}$. Further, we study the uniform exponential mixing of the measure $μ^{*}$ in the sense of Wasserstein metric. Fourthly, under uniform exponential mixing and Markov property, we establish the strong law of large numbers, the central limit theorem and estimate the corresponding rates of convergence for solution maps of SFPDEs. Finally, we give applications of stochastic generalized porous media equations with delay to illustrate of our results.

math.PR↗

arXiv2024

An anti-noise seismic inversion method based on diffusion model

Seismic impedance inversion is one of the most important part of geophysical exploration. However, due to random noise, the traditional semi-supervised learning (SSL) methods lack generalization and stability. To solve this problem, some authors have proposed SSL methods with anti-noise function to improve noise robustness and inversion accuracy. However, such methods are often not ideal when faced with strong noise. In addition, Low-frequency impedance models can mitigate this problem, but creating accurate low-frequency models is difficult and error-prone when well-log data is sparse and subsurface structures are complex. To address those issues, we propose a novel deep learning inversion method called DSIM-USSL (Unsupervised and Semi-supervised joint Learning for Seismic Inversion based on diffusion model). Specifically, we are the first to introduce a diffusion model with strong noise tendency and construct a diffusion seismic inversion model (DSIM). In the reverse diffusion of DSIM, we design the encoder-decoder which combines CNN for capturing local features and GRU for global sequence modeling; and we choose U-net to learn the distribution of random noise, enhancing the generalization and stability of proposed method. Furthermore, to further improve generalization of the proposed method, a two-step training approach (USSL) is utilized. First, an unsupervised trained encoder-decoder is used as the initial network model in place of the traditional low-frequency wave impedance model that is difficult to accurately acquire. Then, the SSL is employed to further optimize the encoder-decoder model. Experimental results on the Marmousi2 model and field data demonstrate that the DSIM-USSL method achieves higher accuracy in the presence of seismic data with random noise, and maintains high stability even under strong noise conditions.

physics.geo-ph↗

arXiv2024

Stance Detection with Collaborative Role-Infused LLM-Based Agents

Stance detection automatically detects the stance in a text towards a target, vital for content analysis in web and social media research. Despite their promising capabilities, LLMs encounter challenges when directly applied to stance detection. First, stance detection demands multi-aspect knowledge, from deciphering event-related terminologies to understanding the expression styles in social media platforms. Second, stance detection requires advanced reasoning to infer authors' implicit viewpoints, as stance are often subtly embedded rather than overtly stated in the text. To address these challenges, we design a three-stage framework COLA (short for Collaborative rOle-infused LLM-based Agents) in which LLMs are designated distinct roles, creating a collaborative system where each role contributes uniquely. Initially, in the multidimensional text analysis stage, we configure the LLMs to act as a linguistic expert, a domain specialist, and a social media veteran to get a multifaceted analysis of texts, thus overcoming the first challenge. Next, in the reasoning-enhanced debating stage, for each potential stance, we designate a specific LLM-based agent to advocate for it, guiding the LLM to detect logical connections between text features and stance, tackling the second challenge. Finally, in the stance conclusion stage, a final decision maker agent consolidates prior insights to determine the stance. Our approach avoids extra annotated data and model training and is highly usable. We achieve state-of-the-art performance across multiple datasets. Ablation studies validate the effectiveness of each design role in handling stance detection. Further experiments have demonstrated the explainability and the versatility of our approach. Our approach excels in usability, accuracy, effectiveness, explainability and versatility, highlighting its value.

cs.CL↗

arXiv2024

Amplifying Frequency Up-Converted Infrared Signals with a Molecular Optomechanical Cavity

Frequency up-conversion, enabled by molecular optomechanical coupling, has recently emerged as a promising approach for converting infrared signals into the visible range through quantum coherent conversion of signals. However, detecting these converted signals poses a significant challenge due to their inherently weak signal intensity. In this work, we propose an amplification mechanism capable of enhancing the signal intensity by a factor of 1000 or more for the frequency up-converted infrared signal in a molecular optomechanical system. The mechanism takes advantage of the strong coupling enhancement with molecular collective mode and Stokes sideband pump. This work demonstrates a feasible approach for up-converting infrared signals to the visible range.

quant-ph↗

arXiv2024

Understanding Expressivity of GNN in Rule Learning

Rule learning is critical to improving knowledge graph (KG) reasoning due to their ability to provide logical and interpretable explanations. Recently, Graph Neural Networks (GNNs) with tail entity scoring achieve the state-of-the-art performance on KG reasoning. However, the theoretical understandings for these GNNs are either lacking or focusing on single-relational graphs, leaving what the kind of rules these GNNs can learn an open problem. We propose to fill the above gap in this paper. Specifically, GNNs with tail entity scoring are unified into a common framework. Then, we analyze their expressivity by formally describing the rule structures they can learn and theoretically demonstrating their superiority. These results further inspire us to propose a novel labeling strategy to learn more rules in KG reasoning. Experimental results are consistent with our theoretical findings and verify the effectiveness of our proposed method. The code is publicly available at https://github.com/LARS-research/Rule-learning-expressivity.

cs.LG↗