arXiv subjects

Yan Leng

Publications and source records attributed to Yan Leng.

At least 19 recordsLinked to original sources

arXiv2026

IDEEA: training-free Input-Dependent stEEring via Activation cluster matching

Steering aligns large language models (LLMs) by injecting a bias into selected activations at inference time, offering a far cheaper alternative to weight-update methods such as supervised fine-tuning or reinforcement learning. However, most existing training-free steering methods are input-independent: a single direction is fitted once and shared across all inputs. This is fundamentally limiting as different inputs occupy different regions of the activation space and admit different optimal steering directions toward the same target concept, much as the gradient with respect to a fixed loss varies from input to input. We close this gap with IDEEA (Input-Dependent stEEring via Activation cluster matching), a training-free framework for input-dependent steering. IDEEA clusters the positive and negative activation supports per attention head, and solves an optimal-matching problem to construct a set of cluster-conditional directions, all about the target concept. At inference time, it picks from this pool of directions and uses the one that best matches the input's own activation for steering. IDEEA aligns the model toward the target concept while preserving the input's original representation, evidence that activations encoding a concept occupy several distinct sub-regions of the representation space rather than a single one. IDEEA improves the truth $\times$ info rate in TruthfulQA by an average of 9.9% (up to 23.5%) over the best input-independent baseline.

cs.CL↗

arXiv2026

Deep ZakaiJ: Structured Filtering for Jump-Diffusion Time Series Forecasting

Time series driven by unobserved latent states frequently exhibit abrupt jump discontinuities whose timing and magnitude cannot be predicted from observed history alone. Classical jump-diffusion models offer a principled mathematical framework but assume rigid parametric forms, while recent neural jump models operate on fully observed trajectories without inferring the hidden states that govern the dynamics. We propose \textit{Deep ZakaiJ}, a latent-state model for partially observed jump-diffusion systems that embeds the Zakai nonlinear filtering equation into a neural encoder--decoder architecture. The encoder recursively updates a belief over the latent state via Strang splitting into three interpretable substeps: prior propagation, diffusion innovation, and jump innovation, yielding a differentiable, first-order-accurate approximation of the exact filtering evolution. The decoder is a structured jump-diffusion model explicitly conditioned on the filtered belief, preserving the separation between continuous dynamics and discontinuous shocks. On synthetic, financial, and oceanographic datasets, \textit{Deep ZakaiJ} improves distributional forecasts while remaining competitive in point accuracy, achieving calibrated predictive intervals and recovering interpretable latent structure in synthetic and qualitative case studies.

cs.LG↗

arXiv2025

Natural Language Processing for Cardiology: A Narrative Review

Cardiovascular diseases are becoming increasingly prevalent in modern society, with a profound impact on global health and well-being. These Cardiovascular disorders are complex and multifactorial, influenced by genetic predispositions, lifestyle choices, and diverse socioeconomic and clinical factors. Information about these interrelated factors is dispersed across multiple types of textual data, including patient narratives, medical records, and scientific literature. Natural language processing (NLP) has emerged as a powerful approach for analysing such unstructured data, enabling healthcare professionals and researchers to gain deeper insights that may transform the diagnosis, treatment, and prevention of cardiac disorders. This review provides a comprehensive overview of NLP research in cardiology from 2014 to 2025. We systematically searched six literature databases for studies describing NLP applications across a range of cardiovascular diseases. After a rigorous screening process, we identified 265 relevant articles. Each study was analysed across multiple dimensions, including NLP paradigms, cardiology-related tasks, disease types, and data sources. Our findings reveal substantial diversity within these dimensions, reflecting the breadth and evolution of NLP research in cardiology. A temporal analysis further highlights methodological trends, showing a progression from rule-based systems to large language models. Finally, we discuss key challenges and future directions, such as developing interpretable LLMs and integrating multimodal data. To the best of our knowledge, this review represents the most comprehensive synthesis of NLP research in cardiology to date.

cs.CL↗

arXiv2025

Demo: Guide-RAG: Evidence-Driven Corpus Curation for Retrieval-Augmented Generation in Long COVID

As AI chatbots gain adoption in clinical medicine, developing effective frameworks for complex, emerging diseases presents significant challenges. We developed and evaluated six Retrieval-Augmented Generation (RAG) corpus configurations for Long COVID (LC) clinical question answering, ranging from expert-curated sources to large-scale literature databases. Our evaluation employed an LLM-as-a-judge framework across faithfulness, relevance, and comprehensiveness metrics using LongCOVID-CQ, a novel dataset of expert-generated clinical questions. Our RAG corpus configuration combining clinical guidelines with high-quality systematic reviews consistently outperformed both narrow single-guideline approaches and large-scale literature databases. Our findings suggest that for emerging diseases, retrieval grounded in curated secondary reviews provides an optimal balance between narrow consensus documents and unfiltered primary literature, supporting clinical decision-making while avoiding information overload and oversimplified guidance. We propose Guide-RAG, a chatbot system and accompanying evaluation framework that integrates both curated expert knowledge and comprehensive literature databases to effectively answer LC clinical questions.

cs.AI↗

arXiv2025

Neural MJD: Neural Non-Stationary Merton Jump Diffusion for Time Series Prediction

While deep learning methods have achieved strong performance in time series prediction, their black-box nature and inability to explicitly model underlying stochastic processes often limit their generalization to non-stationary data, especially in the presence of abrupt changes. In this work, we introduce Neural MJD, a neural network based non-stationary Merton jump diffusion (MJD) model. Our model explicitly formulates forecasting as a stochastic differential equation (SDE) simulation problem, combining a time-inhomogeneous It\^o diffusion to capture non-stationary stochastic dynamics with a time-inhomogeneous compound Poisson process to model abrupt jumps. To enable tractable learning, we introduce a likelihood truncation mechanism that caps the number of jumps within small time intervals and provide a theoretical error bound for this approximation. Additionally, we propose an Euler-Maruyama with restart solver, which achieves a provably lower error bound in estimating expected states and reduced variance compared to the standard solver. Experiments on both synthetic and real-world datasets demonstrate that Neural MJD consistently outperforms state-of-the-art deep learning and statistical learning methods.

cs.LG↗

arXiv2025

Adapting to LLMs: How Insiders and Outsiders Reshape Scientific Knowledge Production

CSCW has long examined how emerging technologies reshape the ways researchers collaborate and produce knowledge, with scientific knowledge production as a central area of focus. As AI becomes increasingly integrated into scientific research, understanding how researchers adapt to it reveals timely opportunities for CSCW research -- particularly in supporting new forms of collaboration, knowledge practices, and infrastructure in AI-driven science. This study quantifies LLM impacts on scientific knowledge production based on an evaluation workflow that combines an insider-outsider perspective with a knowledge production framework. Our findings reveal how LLMs catalyze both innovation and reorganization in scientific communities, offering insights into the broader transformation of knowledge production in the age of generative AI and sheds light on new research opportunities in CSCW.

cs.HC↗

arXiv2025

TAMA: A Human-AI Collaborative Thematic Analysis Framework Using Multi-Agent LLMs for Clinical Interviews

Thematic analysis (TA) is a widely used qualitative approach for uncovering latent meanings in unstructured text data. TA provides valuable insights in healthcare but is resource-intensive. Large Language Models (LLMs) have been introduced to perform TA, yet their applications in high-stakes healthcare settings, particularly for qualitative clinical interview analysis, remain limited. Here, we propose TAMA: A Human-AI Collaborative Thematic Analysis framework using Multi-Agent LLMs for clinical interviews. We leverage the scalability and coherence of multi-agent systems through structured conversations between agents and coordinate the expertise of cardiac experts in TA. Using interview transcripts from parents of children with Anomalous Aortic Origin of a Coronary Artery (AAOCA), a rare congenital heart disease, we demonstrate that TAMA outperforms single-agent LLM TA approaches, achieving higher thematic hit rate, coverage, and distinctiveness. TAMA demonstrates strong potential for automated TA in clinical settings by leveraging multi-agent LLM systems with human-in-the-loop integration by enhancing quality while significantly reducing manual workload. The full implementation is publicly available at https://github.com/Charlie-Yi-SJ/TAMA.

cs.HC↗

arXiv2025

Toward a Flexible Framework for Linear Representation Hypothesis Using Maximum Likelihood Estimation

Linear representation hypothesis posits that high-level concepts are encoded as linear directions in the representation spaces of LLMs. Park et al. (2024) formalize this notion by unifying multiple interpretations of linear representation, such as 1-dimensional subspace representation and interventions, using a causal inner product. However, their framework relies on single-token counterfactual pairs and cannot handle ambiguous contrasting pairs, limiting its applicability to complex or context-dependent concepts. We introduce a new notion of binary concepts as unit vectors in a canonical representation space, and utilize LLMs' (neural) activation differences along with maximum likelihood estimation (MLE) to compute concept directions (i.e., steering vectors). Our method, Sum of Activation-base Normalized Difference (SAND), formalizes the use of activation differences modeled as samples from a von Mises-Fisher (vMF) distribution, providing a principled approach to derive concept directions. We extend the applicability of Park et al. (2024) by eliminating the dependency on unembedding representations and single-token pairs. Through experiments with LLaMA models across diverse concepts and benchmarks, we demonstrate that our lightweight approach offers greater flexibility, superior performance in activation engineering tasks like monitoring and manipulation.

cs.LG↗

arXiv2024

FusionTransNet for Smart Urban Mobility: Spatiotemporal Traffic Forecasting Through Multimodal Network Integration

This study develops FusionTransNet, a framework designed for Origin-Destination (OD) flow predictions within smart and multimodal urban transportation systems. Urban transportation complexity arises from the spatiotemporal interactions among various traffic modes. Motivated by analyzing multimodal data from Shenzhen, a framework that can dissect complicated spatiotemporal interactions between these modes, from the microscopic local level to the macroscopic city-wide perspective, is essential. The framework contains three core components: the Intra-modal Learning Module, the Inter-modal Learning Module, and the Prediction Decoder. The Intra-modal Learning Module is designed to analyze spatial dependencies within individual transportation modes, facilitating a granular understanding of single-mode spatiotemporal dynamics. The Inter-modal Learning Module extends this analysis, integrating data across different modes to uncover cross-modal interdependencies, by breaking down the interactions at both local and global scales. Finally, the Prediction Decoder synthesizes insights from the preceding modules to generate accurate OD flow predictions, translating complex multimodal interactions into forecasts. Empirical evaluations conducted in metropolitan contexts, including Shenzhen and New York, demonstrate FusionTransNet's superior predictive accuracy compared to existing state-of-the-art methods. The implication of this study extends beyond urban transportation, as the method for transferring information across different spatiotemporal graphs at both local and global scales can be instrumental in other spatial systems, such as supply chain logistics and epidemics spreading.

cs.LG↗

arXiv2023

Do LLM Agents Exhibit Social Behavior?

As LLMs increasingly take on roles in human-AI interactions and autonomous AI systems, understanding their social behavior becomes important for informed use and continuous improvement. However, their behaviors in social interactions with humans and other agents, as well as the mechanisms shaping their responses, remain underexplored. To address this gap, we introduce a novel probabilistic framework, State-Understanding-Value-Action (SUVA), to systematically analyze LLM responses in social contexts based on their textual outputs (i.e., utterances). Using canonical behavioral economics games and social preference concepts relatable to LLM users, SUVA assesses LLMs' social behavior through both their final decisions and the response generation processes leading to those decisions. Our analysis of eight LLMs -- including two GPT, four LLaMA, and two Mistral models -- suggests that most models do not generate decisions aligned solely with self-interest; instead, they often produce responses that reflect social welfare considerations and display patterns consistent with direct and indirect reciprocity. Additionally, higher-capacity models more frequently display group identity effects. The SUVA framework also provides explainable tools -- including tree-based visualizations and probabilistic dependency analysis -- to elucidate how factors in LLMs' utterance-based reasoning influence their decisions. We demonstrate that utterance-based reasoning reliably predicts LLMs' final actions; references to altruism, fairness, and cooperation in the reasoning increase the likelihood of prosocial actions, while mentions of self-interest and competition reduce them. Overall, our framework enables practitioners to assess LLMs for applications involving social interactions, and provides researchers with a structured method to interpret how LLM behavior arises from utterance-based reasoning.

cs.AI↗

arXiv2022

Learning to Infer Structures of Network Games

Strategic interactions between a group of individuals or organisations can be modelled as games played on networks, where a player's payoff depends not only on their actions but also on those of their neighbours. Inferring the network structure from observed game outcomes (equilibrium actions) is an important problem with numerous potential applications in economics and social sciences. Existing methods mostly require the knowledge of the utility function associated with the game, which is often unrealistic to obtain in real-world scenarios. We adopt a transformer-like architecture which correctly accounts for the symmetries of the problem and learns a mapping from the equilibrium actions to the network structure of the game without explicit knowledge of the utility function. We test our method on three different types of network games using both synthetic and real-world data, and demonstrate its effectiveness in network structure inference and superior performance over existing methods.

cs.LG↗

arXiv2020

Pandemics are catalysts of scientific novelty: Evidence from COVID-19

Scientific novelty drives the efforts to invent new vaccines and solutions during the pandemic. First-time collaboration and international collaboration are two pivotal channels to expand teams' search activities for a broader scope of resources required to address the global challenge, which might facilitate the generation of novel ideas. Our analysis of 98,981 coronavirus papers suggests that scientific novelty measured by the BioBERT model that is pre-trained on 29 million PubMed articles, and first-time collaboration increased after the outbreak of COVID-19, and international collaboration witnessed a sudden decrease. During COVID-19, papers with more first-time collaboration were found to be more novel and international collaboration did not hamper novelty as it had done in the normal periods. The findings suggest the necessity of reaching out for distant resources and the importance of maintaining a collaborative scientific community beyond nationalism during a pandemic.

cs.DL↗

arXiv2020

Social Learning and the Accuracy-Risk Trade-off in the Wisdom of the Crowd

How do we design and deploy crowdsourced prediction platforms for real-world applications where risk is an important dimension of prediction performance? To answer this question, we conducted a large online Wisdom of the Crowd study where participants predicted the prices of real financial assets (e.g. S&P 500). We observe a Pareto frontier between accuracy of prediction and risk, and find that this trade-off is mediated by social learning i.e. as social learning is increasingly leveraged, it leads to lower accuracy but also lower risk. We also observe that social learning leads to superior accuracy during one of our rounds that occurred during the high market uncertainty of the Brexit vote. Our results have implications for the design of crowdsourced prediction platforms: for example, they suggest that the performance of the crowd should be more comprehensively characterized by using both accuracy and risk (as is standard in financial and statistical forecasting), in contrast to prior work where risk of prediction has been overlooked.

stat.AP↗

arXiv2020

Interpretable Stochastic Block Influence Model: measuring social influence among homophilous communities

Decision-making on networks can be explained by both homophily and social influence. While homophily drives the formation of communities with similar characteristics, social influence occurs both within and between communities. Social influence can be reasoned through role theory, which indicates that the influences among individuals depend on their roles and the behavior of interest. To operationalize these social science theories, we empirically identify the homophilous communities and use the community structures to capture the "roles", which affect the particular decision-making processes. We propose a generative model named Stochastic Block Influence Model and jointly analyze both the network formation and the behavioral influence within and between different empirically-identified communities. To evaluate the performance and demonstrate the interpretability of our method, we study the adoption decisions of microfinance in an Indian village. We show that although individuals tend to form links within communities, there are strong positive and negative social influences between communities, supporting the weak tie theory. Moreover, we find that communities with shared characteristics are associated with positive influence. In contrast, the communities with a lack of overlap are associated with negative influence. Our framework facilitates the quantification of the influences underlying decision communities and is thus a useful tool for driving information diffusion, viral marketing, and technology adoptions.

cs.SI↗

arXiv2020

Analysis of misinformation during the COVID-19 outbreak in China: cultural, social and political entanglements

COVID-19 resulted in an infodemic, which could erode public trust, impede virus containment, and outlive the pandemic itself. The evolving and fragmented media landscape is a key driver of the spread of misinformation. Using misinformation identified by the fact-checking platform by Tencent and posts on Weibo, our results showed that the evolution of misinformation follows an issue-attention cycle, pertaining to topics such as city lockdown, cures, and preventions, and school reopening. Sources of authority weigh in on these topics, but their influence is complicated by peoples' pre-existing beliefs and cultural practices. Finally, social media has a complicated relationship with established or legacy media systems. Sometimes they reinforce each other, but in general, social media may have a topic cycle of its own making. Our findings shed light on the distinct characteristics of misinformation during the COVID-19 and offer insights into combating misinformation in China and across the world at large.

cs.SI↗

arXiv2018

Learning Quadratic Games on Networks

Individuals, or organizations, cooperate with or compete against one another in a wide range of practical situations. Such strategic interactions are often modeled as games played on networks, where an individual's payoff depends not only on her action but also on that of her neighbors. The current literature has largely focused on analyzing the characteristics of network games in the scenario where the structure of the network, which is represented by a graph, is known beforehand. It is often the case, however, that the actions of the players are readily observable while the underlying interaction network remains hidden. In this paper, we propose two novel frameworks for learning, from the observations on individual actions, network games with linear-quadratic payoffs, and in particular, the structure of the interaction network. Our frameworks are based on the Nash equilibrium of such games and involve solving a joint optimization problem for the graph structure and the individual marginal benefits. Both synthetic and real-world experiments demonstrate the effectiveness of the proposed frameworks, which have theoretical as well as practical implications for understanding strategic interactions in a network environment.

cs.GT↗

arXiv2018

Contextual Centrality: Going Beyond Network Structures

Centrality is a fundamental network property which ranks nodes by their structural importance. However, structural importance may not suffice to predict successful diffusions in a wide range of applications, such as word-of-mouth marketing and political campaigns. In particular, nodes with high structural importance may contribute negatively to the objective of the diffusion. To address this problem, we propose contextual centrality, which integrates structural positions, the diffusion process, and, most importantly, nodal contributions to the objective of the diffusion. We perform an empirical analysis of the adoption of microfinance in Indian villages and weather insurance in Chinese villages. Results show that contextual centrality of the first-informed individuals has higher predictive power towards the eventual adoption outcomes than other standard centrality measures. Interestingly, when the product of diffusion rate $p$ and the largest eigenvalue $\lambda_1$ is larger than one and diffusion period is long, contextual centrality linearly scales with eigenvector centrality. This approximation reveals that contextual centrality identifies scenarios where a higher diffusion rate of individuals may negatively influence the cascade payoff. Further simulations on the synthetic and real-world networks show that contextual centrality has the advantage of selecting an individual whose local neighborhood generates a high cascade payoff when $p \lambda_1 < 1$. Under this condition, stronger homophily leads to higher cascade payoff. Our results suggest that contextual centrality captures more complicated dynamics on networks and has significant implications for applications, such as information diffusion, viral marketing, and political campaigns.

cs.SI↗

arXiv2018

Profiling presence patterns and segmenting user locations from cell phone data

The dynamic monitoring of commuting flows is crucial for improving transit systems in fast-developing cities around the world. However, existing methodology to infer commuting originations and destinations have to either rely on large-scale survey data, which is inherently expensive to implement, or on Call Detail Records but based on ad-hoc heuristic assignment rules based on the frequency of appearance at given locations. In this paper, we proposed a novel method to accurately infer the point of origin and destinations of commuting flows based on individual's spatial-temporal patterns inferred from Call Detail Records. Our project significantly improves the accuracy upon the heuristic assignment rules popularly adopted in the literature. Starting with the historical data of geo-temporal travel patterns for a panel of individuals, we create, for each person-location, a vector of probability distribution capturing the likelihood that the person will appear in that location for a given the time of day. Stacked in this way, the matrix of historical geo-temporal data enables us to apply Eigen-decomposition and use unsupervised machine learning techniques to extract commonalities across locations for the different groups of travelers, which ultimately allows us to make inferences and create labels, such as home and work, on specific locations. Testing the methodology on real-world data with known location labels shows that our method identifies home and workplaces with significant accuracy, improving upon the most commonly used methods in the literature by 79% and 34%, respectively. Most importantly, our methodology does not bear any significant computation burden and is easily scalable and easily expanded to other real-world data with historical tracking.

cs.SI↗