arXiv subjects

Yining Liu

Publications and source records attributed to Yining Liu.

17 recordsLinked to original sources

arXiv2026

GRACE: Grounded Reasoning via Adapter Composition and Evidence-Aware Calibration for Educational Visual Question Answering

Educational visual question answering, or VQA, requires models to solve curriculum-oriented multiple-choice questions using both language and visual evidence. Compared with conventional open-ended VQA, educational examples often include structured assessment metadata, diagrams or image contexts, and semantically close answer options, creating strong opportunities for question-option shortcuts. We develop and evaluate a parameter-efficient adaptation framework for a frozen multimodal large language model in this setting. We introduce GRACE, Grounded Reasoning via Adapter Composition and Evidence-Aware Calibration, a framework that uses the pedagogical state of each question to specialize lightweight language and vision adaptation. The state combines inference-visible subject, grouped skill, grade, visual-context, question-intent, and option-structure cues. GRACE uses factor-specific prompts and lightweight visual adapters, then applies evidence-aware option calibration to score all candidates under a shared multimodal context. On ScienceQA, GRACE improves a shared-adapter baseline from 90.5 percent to 93.1 percent overall accuracy and from 88.7 percent to 91.2 percent on image-context questions. Removing pedagogical composition, option calibration, or the visual adapter reduces overall accuracy by 1.4, 1.0, and 1.5 points, respectively. These controlled results show that structured educational state is an effective routing signal for parameter-efficient multimodal adaptation.

cs.MM↗

arXiv2026

MIRA: Evidence-Verified Repair Memory for Text-to-SQL Correction

Text-to-SQL agents still produce executable yet semantically incorrect SQL. A reliable SQL corrector must repair incorrect queries without corrupting correct ones. Confirmed corrections from the same database can be reused without parameter updates. Existing methods, however, often bundle multiple errors and their repairs into a single coarse-grained experience. Applying the entire experience can introduce irrelevant edits and turn an initially correct query into an incorrect one. Reliable reuse therefore depends on three decisions: what to retain from a historical correction, when to activate the resulting memory, and how to adapt it to the current SQL. We propose MIRA (Memory-Item Reuse and Adaptation), a pluggable SQL corrector that uses database evidence to guide memory reuse. MIRA converts historical corrections into independently reusable repair memory items. For each current query, it retrieves memory items using the question and SQL. It then checks each item against database evidence and adapts the supported items to the current SQL. We evaluate 1,785 test queries generated by three Text-to-SQL agents across 14 databases from BIRD and ScienceBenchmark. MIRA improves execution accuracy by 16.53% and 8.78% on BIRD and ScienceBenchmark, respectively.

cs.DB↗

arXiv2026

RECAST: A Region-Scoped Adaptive Index for Exact Similarity Search

Similarity search in metric spaces is widely used in bioinformatics, data mining, and recommender systems. Exact similarity search is dominated by distance computations, while real query streams often concentrate in specific regions rather than spreading uniformly across the space. Pre-built indexes are constructed before the query stream and cannot adapt when queries concentrate in poorly served regions. Adaptive indexes such as AV-tree build the index from distances computed while answering queries, but discard many of those distances and do not organize the retained distances effectively for reuse. We propose RECAST, a region-scoped adaptive index for exact similarity search. RECAST maintains query regions, accumulates distances already computed while answering earlier queries (paid distances) within each region for exact pruning, and uses changes in their pruning effect to infer whether incoming queries remain concentrated in the current region. When queries shift, RECAST recursively dispatches query work to child regions, so paid distances are accumulated and reused only where they remain effective. On five real-world datasets under four workload patterns, RECAST achieves consistently lower cumulative cost than the adaptive baseline and most pre-built baselines, reducing per-query distance computations by up to 64% and query time by up to 46% compared with the state-of-the-art adaptive baseline AV-tree.

cs.DB↗

arXiv2026

Secure Long-Range Autonomous Valet Parking: A Reservation Scheme With Three-Factor Authentication and Key Agreement

Long-range autonomous valet parking (LAVP) is increasingly adopted to alleviate traffic congestion and parking difficulties. For large-scale parking demand, reservation can improve parking management. However, existing schemes mainly focus on parking request verification and parking check-in, and do not adequately protect identity legitimacy and communication security during passenger drop-off and pick-up. To address this problem, we propose SecLAVP, a provably secure three-factor authentication and key agreement protocol for LAVP reservation services. SecLAVP combines passwords, biometrics, and smart cards. With assistance from the drop-off/pick-up point (DP), the passenger and the autonomous vehicle (AV) achieve mutual authentication and establish a session key for secure communication. In the Real-Or-Random (ROR) model, we formally prove that SecLAVP provides session-key security. AVISPA simulations show that SecLAVP resists man-in-the-middle attacks, while informal analysis demonstrates that it satisfies 15 defined security goals. Finally, performance evaluation in terms of communication overhead, computational overhead, and scheduling shows that SecLAVP is feasible for practical deployment.

cs.CR↗

arXiv2026

MIS-HCC: Hierarchical Channel Clustering for Efficient Medical Image Segmentation

Medical image segmentation models require both high accuracy and lightweight design to accommodate real-world medical applications. The deployment of these models on resource-limited medical platforms remains a significant challenge due to their high computational and parameter requirements. Existing pruning methods for model compression mostly overlook the intrinsic connections and similarity between the internal structures of complex deep neural networks. As a result, compressed models may not effectively retain the basic features of the pretrained network. To solve this problem, we propose a hierarchical clustering compression method for medical image segmentation models (MIS-HCC). This approach employs hierarchical clustering to partition channels and fuse their parameters efficiently. Specifically, it leverages the Wasserstein distance to represent similarity of channels within layers of pre-trained network, forming a similarity matrix that guides the clustering process. Channels within each cluster are then fused to produce a compressed network. Experimental results on three medical image datasets application demonstrate that MIS-HCC outperforms the state-of-the-art methods in both accuracy and compression efficiency, offering an effective solution for deploying medical image segmentation models on resource-limited medical platforms.

cs.CV↗

arXiv2026

MoEBlaze: Breaking the Memory Wall for Efficient MoE Training on Modern GPUs

The pervasive "memory wall" bottleneck is significantly amplified in modern large-scale Mixture-of-Experts (MoE) architectures. MoE's inherent architectural sparsity leads to sparse arithmetic compute and also introduces substantial activation memory overheads -- driven by large token routing buffers and the need to materialize and buffer intermediate tensors. This memory pressure limits the maximum batch size and sequence length that can fit on GPUs, and also results in excessive data movements that hinders performance and efficient model scaling. We present MoEBlaze, a memory-efficient MoE training framework that addresses these issues through a co-designed system approach: (i) an end-to-end token dispatch and MoE training method with optimized data structures to eliminate intermediate buffers and activation materializing, and (ii) co-designed kernels with smart activation checkpoint to mitigate memory footprint while simultaneously achieving better performance. We demonstrate that MoEBlaze can achieve over 4x speedups and over 50% memory savings compared to existing MoE frameworks.

cs.LG↗

arXiv2026

PartImageNet++ Dataset: Enhancing Visual Models with High-Quality Part Annotations

To address the scarcity of high-quality part annotations in existing datasets, we introduce PartImageNet++ (PIN++), a dataset that provides detailed part annotations for all categories in ImageNet-1K. With 100 annotated images per category, totaling 100K images, PIN++ represents the most comprehensive dataset covering a diverse range of object categories. Leveraging PIN++, we propose a Multi-scale Part-supervised recognition Model (MPM) for robust classification on ImageNet-1K. We first trained a part segmentation network using PIN++ and used it to generate pseudo part labels for the remaining unannotated images. MPM then integrated a conventional recognition architecture with auxiliary bypass layers, jointly supervised by both pseudo part labels and the original part annotations. Furthermore, we conducted extensive experiments on PIN++, including part segmentation, object segmentation, and few-shot learning, exploring various ways to leverage part annotations in downstream tasks. Experimental results demonstrated that our approach not only enhanced part-based models for robust object recognition but also established strong baselines for multiple downstream tasks, highlighting the potential of part annotations in improving model performance. The dataset and the code are available at https://github.com/LixiaoTHU/PartImageNetPP.

cs.CV↗

arXiv2025

KIT's Low-resource Speech Translation Systems for IWSLT2025: System Enhancement with Synthetic Data and Model Regularization

This paper presents KIT's submissions to the IWSLT 2025 low-resource track. We develop both cascaded systems, consisting of Automatic Speech Recognition (ASR) and Machine Translation (MT) models, and end-to-end (E2E) Speech Translation (ST) systems for three language pairs: Bemba, North Levantine Arabic, and Tunisian Arabic into English. Building upon pre-trained models, we fine-tune our systems with different strategies to utilize resources efficiently. This study further explores system enhancement with synthetic data and model regularization. Specifically, we investigate MT-augmented ST by generating translations from ASR data using MT models. For North Levantine, which lacks parallel ST training data, a system trained solely on synthetic data slightly surpasses the cascaded system trained on real data. We also explore augmentation using text-to-speech models by generating synthetic speech from MT data, demonstrating the benefits of synthetic data in improving both ASR and ST performance for Bemba. Additionally, we apply intra-distillation to enhance model performance. Our experiments show that this approach consistently improves results across ASR, MT, and ST tasks, as well as across different pre-trained models. Finally, we apply Minimum Bayes Risk decoding to combine the cascaded and end-to-end systems, achieving an improvement of approximately 1.5 BLEU points.

cs.CL↗

arXiv2025

3D Cell Oversegmentation Correction via Geo-Wasserstein Divergence

3D cell segmentation methods are often hindered by \emph{oversegmentation}, where a single cell is incorrectly split into multiple fragments. This degrades the final segmentation quality and is notoriously difficult to resolve, as oversegmentation errors often resemble natural gaps between adjacent cells. Our work makes two key contributions. First, for 3D cell segmentation, we are the first work to formulate oversegmentation as a concrete problem and propose a geometric framework to identify and correct these errors. Our approach builds a pre-trained classifier using both 2D geometric and 3D topological features extracted from flawed 3D segmentation results. Second, we introduce a novel metric, Geo-Wasserstein divergence, to quantify changes in 2D geometries. This captures the evolving trends of cell mask shape in a geometry-aware manner. We validate our method through extensive experiments on in-domain plant datasets, including both synthesized and real oversegmented cases, as well as on out-of-domain animal datasets to demonstrate transfer learning performance. An ablation study further highlights the contribution of the Geo-Wasserstein divergence. A clear pipeline is provided for end-users to build pre-trained models to any labeled dataset.

cs.CV↗

arXiv2024

ADBM: Adversarial diffusion bridge model for reliable adversarial purification

Recently Diffusion-based Purification (DiffPure) has been recognized as an effective defense method against adversarial examples. However, we find DiffPure which directly employs the original pre-trained diffusion models for adversarial purification, to be suboptimal. This is due to an inherent trade-off between noise purification performance and data recovery quality. Additionally, the reliability of existing evaluations for DiffPure is questionable, as they rely on weak adaptive attacks. In this work, we propose a novel Adversarial Diffusion Bridge Model, termed ADBM. ADBM directly constructs a reverse bridge from the diffused adversarial data back to its original clean examples, enhancing the purification capabilities of the original diffusion models. Through theoretical analysis and experimental validation across various scenarios, ADBM has proven to be a superior and robust defense mechanism, offering significant promise for practical applications.

cs.LG↗

arXiv2024

Speech Editing -- a Summary

With the rise of video production and social media, speech editing has become crucial for creators to address issues like mispronunciations, missing words, or stuttering in audio recordings. This paper explores text-based speech editing methods that modify audio via text transcripts without manual waveform editing. These approaches ensure edited audio is indistinguishable from the original by altering the mel-spectrogram. Recent advancements, such as context-aware prosody correction and advanced attention mechanisms, have improved speech editing quality. This paper reviews state-of-the-art methods, compares key metrics, and examines widely used datasets. The aim is to highlight ongoing issues and inspire further research and innovation in speech editing.

cs.SD↗

arXiv2024

PartImageNet++ Dataset: Scaling up Part-based Models for Robust Recognition

Deep learning-based object recognition systems can be easily fooled by various adversarial perturbations. One reason for the weak robustness may be that they do not have part-based inductive bias like the human recognition process. Motivated by this, several part-based recognition models have been proposed to improve the adversarial robustness of recognition. However, due to the lack of part annotations, the effectiveness of these methods is only validated on small-scale nonstandard datasets. In this work, we propose PIN++, short for PartImageNet++, a dataset providing high-quality part segmentation annotations for all categories of ImageNet-1K (IN-1K). With these annotations, we build part-based methods directly on the standard IN-1K dataset for robust recognition. Different from previous two-stage part-based models, we propose a Multi-scale Part-supervised Model (MPM), to learn a robust representation with part annotations. Experiments show that MPM yielded better adversarial robustness on the large-scale IN-1K over strong baselines across various attack settings. Furthermore, MPM achieved improved robustness on common corruptions and several out-of-distribution datasets. The dataset, together with these results, enables and encourages researchers to explore the potential of part-based models in more real applications.

cs.CV↗

arXiv2024

Planning Charging Stations and Service Operations of Dockless Electric Micromobility Systems

Dockless electric micro-mobility services (e.g., shared e-scooters and e-bikes) have been increasingly popular in the recent decade, and a variety of charging technologies have emerged for these services. The use of charging stations, to/from which service vehicles are transported by the riders for charging, poses as a promising approach because it reduces the need for dedicated staff or contractors. However, unique challenges also arise, such as how to incentivize riders to drop off vehicles at stations and how to efficiently utilize the vehicles being charged at the stations. This paper focuses on dockless e-scooters as an example and develops a new spatial queuing network model to capture the steady-state scooter service cycles, battery consumption and charging processes, and the associated pricing and management mechanisms. Building upon this model, a system of closed-form equations is formulated and incorporated into a constrained nonlinear program to optimize the deployment of the service fleet, the design of charging stations (i.e., number, location, and capacity), user-based charging price promotions and priorities, and repositioning truck operations (i.e., headway and truck load). The proposed queuing network model is found to match very well with agent-based simulations. It is applied to a series of numerical experiments to draw insights into the optimal designs and the system performance. The numerical results reveal strong advantages of using charging stations for shared dockless electric micro-mobility services as compared to state-of-the-art alternatives. The proposed model can also be used to analyze other micromobility services and other charging approaches.

math.OC↗

arXiv2023

Privacy-Preserving Electricity Theft Detection based on Blockchain

In most electricity theft detection schemes, consumers' power consumption data is directly input into the detection center. Although it is valid in detecting the theft of consumers, the privacy of all consumers is at risk unless the detection center is assumed to be trusted. In fact, it is impractical. Moreover, existing schemes may result in some security problems, such as the collusion attack due to the presence of a trusted third party, and malicious data tampering caused by the system operator (SO) being attacked. Aiming at the problems above, we propose a blockchain-based privacy-preserving electricity theft detection scheme without a third party. Specifically, the proposed scheme uses an improved functional encryption scheme to enable electricity theft detection and load monitoring while preserving consumers' privacy; distributed storage of consumers' data with blockchain to resolve security problems such as data tampering, etc. Meanwhile, we build a long short-term memory network (LSTM) model to perform higher accuracy for electricity theft detection. The proposed scheme is evaluated in a real environment, and the results show that it is more accurate in electricity theft detection within acceptable communication and computational overhead. Our system analysis demonstrates that the proposed scheme can resist various security attacks and preserve consumers' privacy.

cs.CR↗

arXiv2023

Language-Driven Anchors for Zero-Shot Adversarial Robustness

Deep Neural Networks (DNNs) are known to be susceptible to adversarial attacks. Previous researches mainly focus on improving adversarial robustness in the fully supervised setting, leaving the challenging domain of zero-shot adversarial robustness an open question. In this work, we investigate this domain by leveraging the recent advances in large vision-language models, such as CLIP, to introduce zero-shot adversarial robustness to DNNs. We propose LAAT, a Language-driven, Anchor-based Adversarial Training strategy. LAAT utilizes the features of a text encoder for each category as fixed anchors (normalized feature embeddings) for each category, which are then employed for adversarial training. By leveraging the semantic consistency of the text encoders, LAAT aims to enhance the adversarial robustness of the image model on novel categories. However, naively using text encoders leads to poor results. Through analysis, we identified the issue to be the high cosine similarity between text encoders. We then design an expansion algorithm and an alignment cross-entropy loss to alleviate the problem. Our experimental results demonstrated that LAAT significantly improves zero-shot adversarial robustness over state-of-the-art methods. LAAT has the potential to enhance adversarial robustness by large-scale multimodal models, especially when labeled data is unavailable during training.

cs.CV↗

arXiv2022

Planning ride-pooling services with detour restrictions for spatially heterogeneous demand: A multi-zone queuing network approach

This study presents a multi-zone queuing network model for steady-state ride-pooling operations that serve heterogeneous demand, and then builds upon this model to optimize the design of ride-pooling services. Spatial heterogeneity is addressed by partitioning the study region into a set of relatively homogeneous zones, and a set of criteria are imposed to avoid significant detours among matched passengers. A generalized multi-zone queuing network model is then developed to describe how vehicles' states transition within each zone and across neighboring zones, and how passengers are served by idle or partially occupied vehicles. A large system of equations is constructed based on the queuing network model to analytically evaluate steady-state system performance. Then, we formulate a constrained nonlinear program to optimize the design of ride-pooling services, such as zone-level vehicle deployment, vehicle routing paths, and vehicle rebalancing operations. A customized solution approach is also proposed to decompose and solve the optimization problem. The proposed model and solution approach are applied to a hypothetical case and a real-world Chicago case study, so as to demonstrate their applicability and to draw insights. Agent-based simulations are also used to corroborate results from the proposed analytical model. These numerical examples not only reveal interesting insights on how ride-pooling services serve heterogeneous demand, but also highlight the importance of addressing demand heterogeneity when designing ride-pooling services.

math.OC↗

arXiv2019

Synthesis of strain-relaxed Ge-Sn alloys using ion implantation and pulsed laser melting

Ge-Sn alloys with a sufficiently high concentration of Sn is a direct bandgap group IV material. Recently, ion implantation followed by pulsed laser melting has been shown to be a promising method to realize this material due to its high reproducibility and precursor-free process. A Ge-Sn alloy with ~9 at.% Sn was shown to be feasible by this technique. However, the compressive strain, inherently occurring in heterogeneous epitaxy of the film, evidently delays the material from the direct bandgap transition. In this report, an attempt to synthesize a highly-relaxed Ge-Sn alloy will be presented. The idea is to produce a significantly thicker film with a higher implant energy and doses. X-ray reciprocal space mapping confirms that the material is largely-relaxed. The peak Sn concentration of the highest dose sample is 6 at.% as determined by Rutherford backscattering spectrometry. Cross-sectional transmission electron microscopy shows unconventional defects in the film as the mechanism for the strain relaxation. Finally, a photoluminescence (PL) study of the strain-relaxed alloys shows photon emission at a wavelength of 2045 nm, suggesting an active incorporation of Sn concentration of ~6 at.%. The results of this study pave way to produce high quality relaxed GeSn alloy using an industrially scalable method.

physics.app-ph↗