arXiv subjects

Yafeng Yin

Publications and source records attributed to Yafeng Yin.

At least 37 records · Page 2Linked to original sources

arXiv2025

Hierarchical Vision-Language Reasoning for Multimodal Multiple-Choice Question Answering

Multimodal Large Language Models (MLLMs) have demonstrated remarkable multimodal understanding capabilities in Visual Question Answering (VQA) tasks by integrating visual and textual features. However, under the challenging ten-choice question evaluation paradigm, existing methods still exhibit significant limitations when processing PDF documents with complex layouts and lengthy content. Notably, current mainstream models suffer from a strong bias toward English training data, resulting in suboptimal performance for Japanese and other language scenarios. To address these challenges, this paper proposes a novel Japanese PDF document understanding framework that combines multimodal hierarchical reasoning mechanisms with Colqwen-optimized retrieval methods, while innovatively introducing a semantic verification strategy through sub-question decomposition. Experimental results demonstrate that our framework not only significantly enhances the model's deep semantic parsing capability for complex documents, but also exhibits superior robustness in practical application scenarios.

cs.IR↗

arXiv2025

Token Prepending: A Training-Free Approach for Eliciting Better Sentence Embeddings from LLMs

Extracting sentence embeddings from large language models (LLMs) is a promising direction, as LLMs have demonstrated stronger semantic understanding capabilities. Previous studies typically focus on prompt engineering to elicit sentence embeddings from LLMs by prompting the model to encode sentence information into the embedding of the last token. However, LLMs are mostly decoder-only models with causal attention and the earlier tokens in the sentence cannot attend to the latter tokens, resulting in biased encoding of sentence information and cascading effects on the final decoded token. To this end, we propose a novel Token Prepending (TP) technique that prepends each layer's decoded sentence embedding to the beginning of the sentence in the next layer's input, allowing earlier tokens to attend to the complete sentence information under the causal attention mechanism. The proposed TP technique is a plug-and-play and training-free technique, which means it can be seamlessly integrated with various prompt-based sentence embedding methods and autoregressive LLMs. Extensive experiments on various Semantic Textual Similarity (STS) tasks and downstream classification tasks demonstrate that our proposed TP technique can significantly improve the performance of existing prompt-based sentence embedding methods across different LLMs, while incurring negligible additional inference cost.

cs.CL↗

arXiv2025

Multi-Prompting Decoder Helps Better Language Understanding

Recent Pre-trained Language Models (PLMs) usually only provide users with the inference APIs, namely the emerging Model-as-a-Service (MaaS) setting. To adapt MaaS PLMs to downstream tasks without accessing their parameters and gradients, some existing methods focus on the output-side adaptation of PLMs, viewing the PLM as an encoder and then optimizing a task-specific decoder for decoding the output hidden states and class scores of the PLM. Despite the effectiveness of these methods, they only use a single prompt to query PLMs for decoding, leading to a heavy reliance on the quality of the adopted prompt. In this paper, we propose a simple yet effective Multi-Prompting Decoder (MPD) framework for MaaS adaptation. The core idea is to query PLMs with multiple different prompts for each sample, thereby obtaining multiple output hidden states and class scores for subsequent decoding. Such multi-prompting decoding paradigm can simultaneously mitigate reliance on the quality of a single prompt, alleviate the issue of data scarcity under the few-shot setting, and provide richer knowledge extracted from PLMs. Specifically, we propose two decoding strategies: multi-prompting decoding with optimal transport for hidden states and calibrated decoding for class scores. Extensive experiments demonstrate that our method achieves new state-of-the-art results on multiple natural language understanding datasets under the few-shot setting.

cs.CL↗

arXiv2025

Aligning LLM with human travel choices: a persona-based embedding learning approach

The advent of large language models (LLMs) presents new opportunities for travel demand modeling. However, behavioral misalignment between LLMs and humans presents obstacles for the usage of LLMs, and existing alignment methods are frequently inefficient or impractical given the constraints of typical travel demand data. This paper introduces a novel framework for aligning LLMs with human travel choice behavior, tailored to the current travel demand data sources. Our framework uses a persona inference and loading process to condition LLMs with suitable prompts to enhance alignment. The inference step establishes a set of base personas from empirical data, and a learned persona loading function driven by behavioral embeddings guides the loading process. We validate our framework on the Swissmetro mode choice dataset, and the results show that our proposed approach significantly outperformed baseline choice models and LLM-based simulation models in predicting both aggregate mode choice shares and individual choice outcomes. Furthermore, we showcase that our framework can generate insights on population behavior through interpretable parameters. Overall, our research offers a more adaptable, interpretable, and resource-efficient pathway to robust LLM-based travel behavior simulation, paving the way to integrate LLMs into travel demand modeling practice in the future.

cs.AI↗

arXiv2025

Contrastive Prompting Enhances Sentence Embeddings in LLMs through Inference-Time Steering

Extracting sentence embeddings from large language models (LLMs) is a practical direction, as it requires neither additional data nor fine-tuning. Previous studies usually focus on prompt engineering to guide LLMs to encode the core semantic information of the sentence into the embedding of the last token. However, the last token in these methods still encodes an excess of non-essential information, such as stop words, limiting its encoding capacity. To this end, we propose a Contrastive Prompting (CP) method that introduces an extra auxiliary prompt to elicit better sentence embedding. By contrasting with the auxiliary prompt, CP can steer existing prompts to encode the core semantics of the sentence, rather than non-essential information. CP is a plug-and-play inference-time intervention method that can be combined with various prompt-based methods. Extensive experiments on Semantic Textual Similarity (STS) tasks and downstream classification tasks demonstrate that our method can improve the performance of existing prompt-based methods across different LLMs. Our code will be released at https://github.com/zifengcheng/CP.

cs.CL↗

arXiv2025

Graph Network for Sign Language Tasks

Recent advances in sign language research have benefited from CNN-based backbones, which are primarily transferred from traditional computer vision tasks (\eg object identification, image recognition). However, these CNN-based backbones usually excel at extracting features like contours and texture, but may struggle with capturing sign-related features. In fact, sign language tasks require focusing on sign-related regions, including the collaboration between different regions (\eg left hand region and right hand region) and the effective content in a single region. To capture such region-related features, we introduce MixSignGraph, which represents sign sequences as a group of mixed graphs and designs the following three graph modules for feature extraction, \ie Local Sign Graph (LSG) module, Temporal Sign Graph (TSG) module and Hierarchical Sign Graph (HSG) module. Specifically, the LSG module learns the correlation of intra-frame cross-region features within one frame, \ie focusing on spatial features. The TSG module tracks the interaction of inter-frame cross-region features among adjacent frames, \ie focusing on temporal features. The HSG module aggregates the same-region features from different-granularity feature maps of a frame, \ie focusing on hierarchical features. In addition, to further improve the performance of sign language tasks without gloss annotations, we propose a simple yet counter-intuitive Text-driven CTC Pre-training (TCP) method, which generates pseudo gloss labels from text labels for model pre-training. Extensive experiments conducted on current five public sign language datasets demonstrate the superior performance of the proposed model. Notably, our model surpasses the SOTA models on multiple sign language tasks across several datasets, without relying on any additional cues.

cs.CV↗

arXiv2025

Toward LLM-Agent-Based Modeling of Transportation Systems: A Conceptual Framework

In transportation system demand modeling and simulation, agent-based models and microsimulations are current state-of-the-art approaches. However, existing agent-based models still have some limitations on behavioral realism and resource demand that limit their applicability. In this study, leveraging the emerging technology of large language models (LLMs) and LLM-based agents, we propose a general LLM-agent-based modeling framework for transportation systems. We argue that LLM agents not only possess the essential capabilities to function as agents but also offer promising solutions to overcome some limitations of existing agent-based models. Our conceptual framework design closely replicates the decision-making and interaction processes and traits of human travelers within transportation networks, and we demonstrate that the proposed systems can meet critical behavioral criteria for decision-making and learning behaviors using related studies and a demonstrative example of LLM agents' learning and adjustment in the bottleneck setting. Although further refinement of the LLM-agent-based modeling framework is necessary, we believe that this approach has the potential to improve transportation system modeling and simulation.

cs.AI↗

arXiv2025

LLM-ABM for Transportation: Assessing the Potential of LLM Agents in System Analysis

Agent-based modeling approaches represent the state-of-art in modeling travel demand and transportation system dynamics and are valuable tools for transportation planning. However, established agent-based approaches in transportation rely on multi-hierarchical mathematical models to simulate travel behavior, which faces theoretical and practical limitations. The advent of large language models (LLM) provides a new opportunity to refine agent-based modeling in transportation. LLM agents, which have impressive reasoning and planning abilities, can serve as a proxy of human travelers and be integrated into the modeling framework. However, despite evidence of their behavioral soundness, no existing studies have assessed the impact and validity of LLM-agent-based simulations from a system perspective in transportation. This paper aims to address this issue by designing and integrating LLM agents with human-traveler-like characteristics into a simulation of a transportation system and assessing its performance based on existing benchmarks. Using the classical transportation setting of the morning commute, we find that not only do the agents exhibit fine behavioral soundness, but also produce system dynamics that align well with standard benchmarks. Our analysis first verifies the effectiveness and potential of LLM-agent-based modeling for transportation planning on the system level.

cs.MA↗

arXiv2025

Implicit Location-Caption Alignment via Complementary Masking for Weakly-Supervised Dense Video Captioning

Weakly-Supervised Dense Video Captioning (WSDVC) aims to localize and describe all events of interest in a video without requiring annotations of event boundaries. This setting poses a great challenge in accurately locating the temporal location of event, as the relevant supervision is unavailable. Existing methods rely on explicit alignment constraints between event locations and captions, which involve complex event proposal procedures during both training and inference. To tackle this problem, we propose a novel implicit location-caption alignment paradigm by complementary masking, which simplifies the complex event proposal and localization process while maintaining effectiveness. Specifically, our model comprises two components: a dual-mode video captioning module and a mask generation module. The dual-mode video captioning module captures global event information and generates descriptive captions, while the mask generation module generates differentiable positive and negative masks for localizing the events. These masks enable the implicit alignment of event locations and captions by ensuring that captions generated from positively and negatively masked videos are complementary, thereby forming a complete video description. In this way, even under weak supervision, the event location and event caption can be aligned implicitly. Extensive experiments on the public datasets demonstrate that our method outperforms existing weakly-supervised methods and achieves competitive results compared to fully-supervised methods.

cs.CV↗

arXiv2024

A Control Theoretic Approach to Simultaneously Estimate Average Value of Time and Determine Dynamic Price for High-occupancy Toll Lanes

The dynamic pricing problem of a freeway corridor with high-occupancy toll (HOT) lanes was formulated and solved based on a point queue abstraction of the traffic system [Yin and Lou, 2009]. However, existing pricing strategies cannot guarantee that the closed-loop system converges to the optimal state, in which the HOT lanes' capacity is fully utilized but there is no queue on the HOT lanes, and a well-behaved estimation and control method is quite challenging and still elusive. This paper attempts to fill the gap by making three fundamental contributions: (i) to present a simpler formulation of the point queue model based on the new concept of residual capacity, (ii) to propose a simple feedback control theoretic approach to estimate the average value of time and calculate the dynamic price, and (iii) to analytically and numerically prove that the closed-loop system is stable and guaranteed to converge to the optimal state, in either Gaussian or exponential manners.

eess.SY↗

arXiv2024

Multiday User Equilibrium with Strategic Commuters

In the era of connected and automated mobility, commuters will possess strong computation power, enabling them to strategically make sequential travel choices over a planning horizon. This paper investigates the multiday traffic patterns that arise from such decision-making behavior. In doing so, we frame the commute problem as a mean-field Markov game and introduce a novel concept of multiday user equilibrium to capture the steady state of commuters' interactions. The proposed model is general and can be tailored to various travel choices such as route or departure time. We explore a range of properties of the multiday user equilibrium under mild conditions. The study reveals the fingerprint of user inertia on network flow patterns, causing between-day variations even at a steady state. Furthermore, our analysis establishes critical connections between the multiday user equilibrium and conventional Wardrop equilibrium.

math.OC↗

arXiv2024

A Debiased Nearest Neighbors Framework for Multi-Label Text Classification

Multi-Label Text Classification (MLTC) is a practical yet challenging task that involves assigning multiple non-exclusive labels to each document. Previous studies primarily focus on capturing label correlations to assist label prediction by introducing special labeling schemes, designing specific model structures, or adding auxiliary tasks. Recently, the $k$ Nearest Neighbor ($k$NN) framework has shown promise by retrieving labeled samples as references to mine label co-occurrence information in the embedding space. However, two critical biases, namely embedding alignment bias and confidence estimation bias, are often overlooked, adversely affecting prediction performance. In this paper, we introduce a DEbiased Nearest Neighbors (DENN) framework for MLTC, specifically designed to mitigate these biases. To address embedding alignment bias, we propose a debiased contrastive learning strategy, enhancing neighbor consistency on label co-occurrence. For confidence estimation bias, we present a debiased confidence estimation strategy, improving the adaptive combination of predictions from $k$NN and inductive binary classifications. Extensive experiments conducted on four public benchmark datasets (i.e., AAPD, RCV1-V2, Amazon-531, and EUR-LEX57K) showcase the effectiveness of our proposed method. Besides, our method does not introduce any extra parameters.

cs.CL↗

arXiv2023

Controlling Class Layout for Deep Ordinal Classification via Constrained Proxies Learning

For deep ordinal classification, learning a well-structured feature space specific to ordinal classification is helpful to properly capture the ordinal nature among classes. Intuitively, when Euclidean distance metric is used, an ideal ordinal layout in feature space would be that the sample clusters are arranged in class order along a straight line in space. However, enforcing samples to conform to a specific layout in the feature space is a challenging problem. To address this problem, in this paper, we propose a novel Constrained Proxies Learning (CPL) method, which can learn a proxy for each ordinal class and then adjusts the global layout of classes by constraining these proxies. Specifically, we propose two kinds of strategies: hard layout constraint and soft layout constraint. The hard layout constraint is realized by directly controlling the generation of proxies to force them to be placed in a strict linear layout or semicircular layout (i.e., two instantiations of strict ordinal layout). The soft layout constraint is realized by constraining that the proxy layout should always produce unimodal proxy-to-proxies similarity distribution for each proxy (i.e., to be a relaxed ordinal layout). Experiments show that the proposed CPL method outperforms previous deep ordinal classification methods under the same setting of feature extractor.

cs.CV↗

arXiv2022

Learning to Classify Open Intent via Soft Labeling and Manifold Mixup

Open intent classification is a practical yet challenging task in dialogue systems. Its objective is to accurately classify samples of known intents while at the same time detecting those of open (unknown) intents. Existing methods usually use outlier detection algorithms combined with K-class classifier to detect open intents, where K represents the class number of known intents. Different from them, in this paper, we consider another way without using outlier detection algorithms. Specifically, we directly train a (K+1)-class classifier for open intent classification, where the (K+1)-th class represents open intents. To address the challenge that training a (K+1)-class classifier with training samples of only K classes, we propose a deep model based on Soft Labeling and Manifold Mixup (SLMM). In our method, soft labeling is used to reshape the label distribution of the known intent samples, aiming at reducing model's overconfident on known intents. Manifold mixup is used to generate pseudo samples for open intents, aiming at well optimizing the decision boundary of open intents. Experiments on four benchmark datasets demonstrate that our method outperforms previous methods and achieves state-of-the-art performance. All the code and data of this work can be obtained at https://github.com/zifengcheng/SLMM.

cs.CL↗

arXiv2022

Efficient Algorithms for Stochastic Ridepooling Assignment with Mixed Fleets

Ride-pooling, which accommodates multiple passenger requests in a single trip, has the potential to significantly increase fleet utilization in shared mobility platforms. The ride-pooling assignment problem finds optimal co-riders to maximize the total utility or profit on a shareability graph, a hypergraph representing the matching compatibility between available vehicles and pending requests. With mixed fleets due to the introduction of automated or premium vehicles, fleet sizing and relocation decisions should be made before the requests are revealed. Due to the immense size of the underlying shareability graph and demand uncertainty, it is impractical to use exact methods to calculate the optimal trip assignments. Two approximation algorithms for mid-capacity and high-capacity vehicles are proposed in this paper; The respective approximation ratios are $\frac1{p^2}$ and $\frac{e-1}{(2e+o(1)) p \ln p}$, where $p$ is the maximum vehicle capacity plus one. The performance of these algorithms is validated using a mixed autonomy on-demand mobility simulator. These efficient algorithms serve as a stepping stone for a variety of multimodal and multiclass on-demand mobility applications.

math.OC↗

arXiv2022

Real-Time Cross-Fleet Pareto-Improving Truck Platoon Coordination

This paper studies a multi-fleet platoon coordination system in transport networks that deploy hubs to form trucks into platoons. The trucks belong to different fleets that are interested in increasing their profits by platooning across fleets. The profit of each fleet incorporates platooning rewards and costs for waiting at hubs. Each truck has a fixed route and a waiting time budget to spend at the hubs along its route. To ensure that all fleets are willing to participate in the system, we develop a cross-fleet Pareto-improving coordination strategy that guarantees higher fleet profits than a coordination strategy without cross-fleet platoons. By leveraging multiple hubs for platoon formation, the coordination strategy can be implemented in a real-time and distributed fashion while largely reducing the amount of travel information to be shared for system-wide coordination. We evaluate the proposed strategy in a simulation study over the Swedish transportation network. The cross-fleet platooning strategy significantly improves fleets' profits compared with single-fleet platooning, especially the profits from smaller fleets. The cross-fleet platooning strategy also shows strong competitiveness in terms of the system-wide profit compared to the case when a system planner optimizes all fleets' total profit.

eess.SY↗

arXiv2021

Heuristics for Customer-focused Ride-pooling Assignment

Ride-pooling has become an important service option offered by ride-hailing platforms as it serves multiple trip requests in a single ride. By leveraging customer data, connected vehicles, and efficient assignment algorithms, ride-pooling can be a critical instrument to address driver shortages and mitigate the negative externalities of ride-hailing operations. Recent literature has focused on computationally intensive optimization-based methods that maximize system throughput or minimize vehicle miles. However, individual customers may experience substantial service quality degradation due to the consequent waiting and detour time. In contrast, this paper examines heuristic methods for real-time ride-pooling assignments that are highly scalable and easily computable. We propose a restricted subgraph method and compare it with other existing heuristic and optimization-based matching algorithms using a variety of metrics. By fusing multiple sources of trip and network data in New York City, we develop a flexible, agent-based simulation platform to test these strategies on different demand levels and examine how they affect both the customer experience and the ride-hailing platform. Our results find a trade-off among heuristics between throughput and customer matching time. We show that our proposed ride-pooling strategy maintains system performance while limiting trip delays and improving customer experience. This work provides insight for policymakers and ride-hailing operators about the performance of simpler heuristics and raises concerns about prioritizing only specific platform metrics without considering service quality.

eess.SY↗

arXiv2020

Modeling Epidemic Spreading through Public Transit using Time-Varying Encounter Network

Passenger contact in public transit (PT) networks can be a key mediate in the spreading of infectious diseases. This paper proposes a time-varying weighted PT encounter network to model the spreading of infectious diseases through the PT systems. Social activity contacts at both local and global levels are also considered. We select the epidemiological characteristics of coronavirus disease 2019 (COVID-19) as a case study along with smart card data from Singapore to illustrate the model at the metropolitan level. A scalable and lightweight theoretical framework is derived to capture the time-varying and heterogeneous network structures, which enables to solve the problem at the whole population level with low computational costs. Different control policies from both the public health side and the transportation side are evaluated. We find that people's preventative behavior is one of the most effective measures to control the spreading of epidemics. From the transportation side, partial closure of bus routes helps to slow down but cannot fully contain the spreading of epidemics. Identifying "influential passengers" using the smart card data and isolating them at an early stage can also effectively reduce the epidemic spreading.

physics.soc-ph↗