arXiv subjects

Hongyu Jin

Publications and source records attributed to Hongyu Jin.

16 recordsLinked to original sources

arXiv2026

RAIL: Rethinking Auditory Intelligence in Large Audio-Language Models with a CHC-Grounded Benchmark

Humans process rich auditory environments through tightly integrated cognitive capabilities such as audio perception, audio reasoning, and memory. Despite recent progress in large audio-language models (LALMs) across speech understanding and multimodal audio reasoning, current evaluation paradigms remain largely task- or modality-centric, focusing on end performance while overlooking underlying auditory cognitive behaviours. This reveals a fundamental gap between how auditory cognition is understood in humans and how it is evaluated in LALMs, particularly in the lack of frameworks that operationalise cognitive principles beyond task-level metrics to systematically capture model behaviour. In this work, we introduce RAIL, a human-centric evaluation paradigm grounded in the Cattell-Horn-Carroll (CHC) cognitive framework. RAIL formalises auditory cognition into five core capabilities and develop them into structured evaluation tasks that probe how models process, retain, and integrate auditory information. We further construct a cognitively grounded benchmark with principled data curation and human-aligned evaluation protocols. Evaluating 26 state-of-the-art LALMs, we find that current models exhibit highly uneven performance across cognitive abilities. RAIL establishes a new evaluation paradigm that moves beyond task-centric benchmarking toward cognitively grounded assessment of auditory intelligence.

cs.SD↗

arXiv2026

RiskProp: Collision-Anchored Self-Supervised Risk Propagation for Early Accident Anticipation

Accident anticipation aims to predict impending collisions from dashcam videos and trigger early alerts. Existing methods rely on binary supervision with manually annotated "anomaly onset" frames, which are subjective and inconsistent, leading to inaccurate risk estimation. In contrast, we propose RiskProp, a novel collision-anchored self-supervised risk propagation paradigm for early accident anticipation, which removes the need for anomaly onset annotations and leverages only the reliably annotated collision frame. RiskProp models temporal risk evolution through two observation-driven losses: first, since future frames contain more definitive evidence of an impending accident, we introduce a future-frame regularization loss that uses the model's next-frame prediction as a soft target to supervise the current frame, enabling backward propagation of risk signals; second, inspired by the empirical trend of rising risk before accidents, we design an adaptive monotonic constraint to encourage a non-decreasing progression over time. Experiments on CAP and Nexar demonstrate that RiskProp achieves state-of-the-art performance and produces smoother, more discriminative risk curves, improving both early anticipation and interpretability.

cs.CV↗

arXiv2026

Scaling Ambiguity: Augmenting Human Annotation in Speech Emotion Recognition with Audio-Language Models

Speech Emotion Recognition models typically use single categorical labels, overlooking the inherent ambiguity of human emotions. Ambiguous Emotion Recognition addresses this by representing emotions as probability distributions, but progress is limited by unreliable ground-truth distributions inferred from sparse human annotations. This paper explores whether Large Audio-Language Models (ALMs) can mitigate the annotation bottleneck by generating high-quality synthetic annotations. We introduce a framework leveraging ALMs to create Synthetic Perceptual Proxies, augmenting human annotations to improve ground-truth distribution reliability. We validate these proxies through statistical analysis of their alignment with human distributions and evaluate their impact by fine-tuning ALMs with the augmented emotion distributions. Furthermore, to address class imbalance and enable unbiased evaluation, we propose DiME-Aug, a Distribution-aware Multimodal Emotion Augmentation strategy. Experiments on IEMOCAP and MSP-Podcast show that synthetic annotations enhance emotion distribution, especially in low-ambiguity regions where annotation agreement is high. However, benefits diminish for highly ambiguous emotions with greater human disagreement. This work provides the first evidence that ALMs could address annotation scarcity in ambiguous emotion recognition, but highlights the need for more advanced prompting or generation strategies to handle highly ambiguous cases.

eess.AS↗

arXiv2025

DNS in the Time of Curiosity: A Tale of Collaborative User Privacy Protection

The Domain Name System (DNS) is central to all Internet user activity, resolving accessed domain names into Internet Protocol (IP) addresses. As a result, curious DNS resolvers can learn everything about Internet users' interests. Public DNS resolvers are rising in popularity, offering low-latency resolution, high reliability, privacy-preserving policies, and support for encrypted DNS queries. However, client-resolver traffic encryption, increasingly deployed to protect users from eavesdroppers, does not protect users against curious resolvers. Similarly, privacy-preserving policies are based solely on written commitments and do not provide technical safeguards. Although DNS query relay schemes can separate duties to limit data accessible by each entity, they cannot prevent colluding entities from sharing user traffic logs. Thus, a key challenge remains: organizations operating public DNS resolvers, accounting for the majority of DNS resolutions, can potentially collect and analyze massive volumes of Internet user activity data. With DNS infrastructure that cannot be fully trusted, can we safeguard user privacy? We answer positively and advocate for a user-driven approach to reduce exposure to DNS services. We will discuss key ideas of the proposal, which aims to achieve a high level of privacy without sacrificing performance: maintaining low latency, network bandwidth, memory/storage overhead, and computational overhead.

cs.CR↗

arXiv2025

LLUAD: Low-Latency User-Anonymized DNS

The Domain Name System (DNS) is involved in practically all web activity, translating easy-to-remember domain names into Internet Protocol (IP) addresses. Due to its central role on the Internet, DNS exposes user web activity in detail. The privacy challenge is honest-but-curious DNS servers/resolvers providing the translation/lookup service. In particular, with the majority of DNS queries handled by public DNS resolvers, the organizations running them can track, collect, and analyze massive user activity data. Existing solutions that encrypt DNS traffic between clients and resolvers are insufficient, as the resolver itself is the privacy threat. While DNS query relays separate duties among multiple entities, to limit the data accessible by each entity, they cannot prevent colluding entities from sharing user traffic logs. To achieve near-zero-trust DNS privacy compatible with the existing DNS infrastructure, we propose LLUAD: it locally stores a Popularity List, the most popular DNS records, on user devices, formed in a privacy-preserving manner based on user interests. In this way, LLUAD can both improve privacy and reduce access times to web content. The Popularity List is proactively retrieved from a (curious) public server that continually updates and refreshes the records based on user popularity votes, while efficiently broadcasting record updates/changes to adhere to aggressive load-balancing schemes (i.e., name servers actively load-balancing user connections by changing record IP addresses). User votes are anonymized using a novel, efficient, and highly scalable client-driven Voting Mix Network - with packet lengths independent of the number of hops, centrally enforced limit on number of votes cast per user, and robustness against poor client participation - to ensure a geographically relevant and correctly/securely instantiated Popularity List.

cs.CR↗

arXiv2025

Accountable, Scalable and DoS-resilient Secure Vehicular Communication

Paramount to vehicle safety, broadcasted Cooperative Awareness Messages (CAMs) and Decentralized Environmental Notification Messages (DENMs) are pseudonymously authenticated for security and privacy protection, with each node needing to have all incoming messages validated within an expiration deadline. This creates an asymmetry that can be easily exploited by external adversaries to launch a clogging Denial of Service (DoS) attack: each forged VC message forces all neighboring nodes to cryptographically validate it; at increasing rates, easy to generate forged messages gradually exhaust processing resources and severely degrade or deny timely validation of benign CAMs/DENMs. The result can be catastrophic when awareness of neighbor vehicle positions or critical reports are missed. We address this problem making the standardized VC pseudonymous authentication DoS-resilient. We propose efficient cryptographic constructs, which we term message verification facilitators, to prioritize processing resources for verification of potentially valid messages among bogus messages and verify multiple messages based on one signature verification. Any message acceptance is strictly based on public-key based message authentication/verification for accountability, i.e., non-repudiation is not sacrificed, unlike symmetric key based approaches. This further enables drastic misbehavior detection, also exploiting the newly introduced facilitators, based on probabilistic signature verification and cross-checking over multiple facilitators verifying the same message; while maintaining verification latency low even when under attack, trading off modest communication overhead. Our facilitators can also be used for efficient discovery and verification of DENM or any event-driven message, including misbehavior evidence used for our scheme.

cs.CR↗

arXiv2024

Deformable-Heatmap-Segmentation for Automobile Visual Perception

Semantic segmentation of road elements in 2D images is a crucial task in the recognition of some static objects such as lane lines and free space. In this paper, we propose DHSNet,which extracts the objects features with a end-to-end architecture along with a heatmap proposal. Deformable convolutions are also utilized in the proposed network. The DHSNet finely combines low-level feature maps with high-level ones by using upsampling operators as well as downsampling operators in a U-shape manner. Besides, DHSNet also aims to capture static objects of various shapes and scales. We also predict a proposal heatmap to detect the proposal points for more accurate target aiming in the network.

cs.CV↗

arXiv2024

Over-the-Air Runtime Wi-Fi MAC Address Re-randomization

Medium Access Control (MAC) address randomization is a key component for privacy protection in Wi-Fi networks. Current proposals periodically change the mobile device MAC addresses when it disconnects from the Access Point (AP). This way frames cannot be linked across changes, but the mobile device presence is exposed as long as it remains connected: all its communication is trivially linkable by observing the randomized yet same MAC address throughout the connection. Our runtime MAC re-randomization scheme addresses this issue, reducing or eliminating Wi-Fi frames linkability without awaiting for or requiring a disconnection. Our MAC re-randomization is practically 'over-the-air': MAC addresses are re-randomized just before transmission, while the protocol stacks (at the mobile and the AP) maintain locally the original connection MAC addresses - making our MAC layer scheme transparent to upper layers. With an implementation and a set of small-scale experiments with off-the-shelf devices, we show the feasibility of our scheme and the potential towards future deployment.

cs.NI↗

arXiv2020

Security and Privacy in Vehicular Social Networks

We surveyed and presented the state-of-the-art VC systems, security and privacy architectures and technologies, emphasizing on security and privacy challenges and their solutions for P2P interactions in VSNs towards standardization and deployment. We note that beyond safety applications that have drawn a lot of attention in VC systems, there is significant and rising interest in vehicle-to-vehicle interaction for a range of transportation efficiency and infotainment applications, notably LBS as well as a gamut of services by mobile providers. While this enriches the VC systems and the user experience, security and privacy concerns are also intensified. This is especially so, considering (i) the privacy risk from the exposure of the users to the service providers, and (ii) the security risk from the interaction with malicious or selfish and thus misbehaving users or infrastructure. We showed existing solutions can in fact evolve and address the VSN-specific challenges, and improve or even accelerate the adoption of VSN applications.

cs.CR↗

arXiv2020

Resilient Privacy Protection for Location-Based Services through Decentralization

Location-Based Services (LBSs) provide valuable services, with convenient features for mobile users. However, the location and other information disclosed through each query to the LBS erodes user privacy. This is a concern especially because LBS providers can be honest-but-curious, collecting queries and tracking users' whereabouts and infer sensitive user data. This motivated both centralized and decentralized location privacy protection schemes for LBSs: anonymizing and obfuscating LBS queries to not disclose exact information, while still getting useful responses. Decentralized schemes overcome disadvantages of centralized schemes, eliminating anonymizers, and enhancing users' control over sensitive information. However, an insecure decentralized system could create serious risks beyond private information leakage. More so, attacking an improperly designed decentralized LBS privacy protection scheme could be an effective and low-cost step to breach user privacy. We address exactly this problem, by proposing security enhancements for mobile data sharing systems. We protect user privacy while preserving accountability of user activities, leveraging pseudonymous authentication with mainstream cryptography. We show our scheme can be deployed with off-the-shelf devices based on an experimental evaluation of an implementation in a static automotive testbed.

cs.CR↗

arXiv2020

Resilient Collaborative Privacy for Location-Based Services

Location-based Services (LBSs) provide valuable services, with convenient features for users. However, the information disclosed through each request harms user privacy. This is a concern particularly with honest-but-curious LBS servers, which could, by collecting requests, track users and infer additional sensitive user data. This is the motivation of both centralized and decentralized location privacy protection schemes for LBSs: anonymizing and obfuscating LBS queries to not disclose exact information, while still getting useful responses. Decentralized schemes overcome the disadvantages of centralized schemes, eliminating anonymizers and enhancing users' control over sensitive information. However, an insecure decentralized system could pose even more serious security threats than privacy leakage. We address exactly this problem, by proposing security enhancements for mobile data sharing systems. We protect user privacy while preserving accountability of user activities, leveraging pseudonymous authentication with mainstream cryptography. Our design leverages architectures proposed for large scale mobile systems, while it incurs minimal changes to LBS servers as it can be deployed in parallel to the LBS servers. This further motivates the adoption of our design, in order to cater to the needs of privacy-sensitive users. We provide an analysis of security and privacy concerns and countermeasures, as well as a performance evaluation of basic protocol operations showing the practicality of our design.

cs.CR↗

arXiv2020

Proactive Certificate Validation for VANETs

Security and privacy in Vehicular Ad-hoc Networks (VANETs) mandates use of short-lived credentials (pseudonyms) and cryptographic key pairs. This implies significant computational overhead for vehicles, needing to validate often numerous such pseudonyms within a short period. To alleviate such a bottleneck that could even place vehicle safety at risk, we propose a proactive pseudonym validation approach based on Bloom Filters (BFs). We show that our scheme could liberate computational resources for other (safety- and time-critical) operations with reasonable communication overhead without compromising security and privacy.

cs.CR↗

arXiv2020

DoS-resilient Cooperative Beacon Verification for Vehicular Communication Systems

Authenticated safety beacons in Vehicular Communication (VC) systems ensure awareness among neighboring vehicles. However, the verification of beacon signatures introduces significant processing overhead for resource-constrained vehicular On-Board Units (OBUs). Even worse in dense neighborhood or when a clogging Denial of Service (DoS) attack is mounted. The OBU would fail to verify for all received (authentic or fictitious) beacons. This could significantly delay the verifications of authentic beacons or even affect the awareness of neighboring vehicle status. In this paper, we propose an efficient cooperative beacon verification scheme leveraging efficient symmetric key based authentication on top of pseudonymous authentication (based on traditional public key cryptography), providing efficient discovery of authentic beacons among a pool of received authentic and fictitious beacons, and can significantly decrease waiting times of beacons in queue before their validations. We show with simulation results that our scheme can guarantee low waiting times for received beacons even in high neighbor density situations and under DoS attacks, under which a traditional scheme would not be workable.

cs.CR↗

arXiv2020

Scaling VANET Security Through Cooperative Message Verification

VANET security introduces significant processing overhead for resource-constrained On-Board Units (OBUs). Here, we propose a novel scheme that allows secure Vehicular Communication (VC) systems to scale well beyond network densities for which existing optimization approaches could be workable, without compromising security (and privacy).

cs.CR↗

arXiv2017

SECMACE: Scalable and Robust Identity and Credential Management Infrastructure in Vehicular Communication Systems

Several years of academic and industrial research efforts have converged to a common understanding on fundamental security building blocks for the upcoming Vehicular Communication (VC) systems. There is a growing consensus towards deploying a special-purpose identity and credential management infrastructure, i.e., a Vehicular Public-Key Infrastructure (VPKI), enabling pseudonymous authentication, with standardization efforts towards that direction. In spite of the progress made by standardization bodies (IEEE 1609.2 and ETSI) and harmonization efforts (Car2Car Communication Consortium (C2C-CC)), significant questions remain unanswered towards deploying a VPKI. Deep understanding of the VPKI, a central building block of secure and privacy-preserving VC systems, is still lacking. This paper contributes to the closing of this gap. We present SECMACE, a VPKI system, which is compatible with the IEEE 1609.2 and ETSI standards specifications. We provide a detailed description of our state-of-the-art VPKI that improves upon existing proposals in terms of security and privacy protection, and efficiency. SECMACE facilitates multi-domain operations in the VC systems and enhances user privacy, notably preventing linking pseudonyms based on timing information and offering increased protection even against honest-but-curious VPKI entities. We propose multiple policies for the vehicle-VPKI interactions, based on which and two large-scale mobility trace datasets, we evaluate the full-blown implementation of SECMACE. With very little attention on the VPKI performance thus far, our results reveal that modest computing resources can support a large area of vehicles with very low delays and the most promising policy in terms of privacy protection can be supported with moderate overhead.

cs.CR↗

arXiv2016

Towards Deploying a Scalable & Robust Vehicular Identity and Credential Management Infrastructure

Several years of academic and industrial research efforts have converged to a common understanding on fundamental security building blocks for the upcoming Vehicular Communication (VC) systems. There is a growing consensus towards deploying a Vehicular Public-Key Infrastructure (VPKI) enables pseudonymous authentication, with standardization efforts in that direction. However, there are still significant technical issues that remain unresolved. Existing proposals for instantiating the VPKI either need additional detailed specifications or enhanced security and privacy features. Equally important, there is limited experimental work that establishes the VPKI efficiency and scalability. In this paper, we are concerned with exactly these issues. We leverage the common VPKI approach and contribute an enhanced system with precisely defined, novel features that improve its resilience and the user privacy protection. In particular, we depart from the common assumption that the VPKI entities are fully trusted and we improve user privacy in the face of an honest-but-curious security infrastructure. Moreover, we fully implement our VPKI, in a standard-compliant manner, and we perform an extensive evaluation. Along with stronger protection and richer functionality, our system achieves very significant performance improvement over prior systems - contributing the most advanced VPKI towards deployment.

cs.CR↗