Projects

Our research spans network, mobile, software, hardware, and human factors, with a consistent emphasis on the security and privacy of these targets. Below we organize our work into major themes, each linked to representative peer-reviewed papers (top-tier and selected venues), newest first. For the complete list, see Papers per year, Conference Papers, and Journal Papers.

AI / Machine-Learning Security

Attacks and defenses for machine-learning algorithms and foundation models — including data poisoning, adversarial evasion, and deepfakes.

  • Cross-Lingual Data Poisoning against Text-to-Image Models — multilingual vulnerability factors and defenses (ICPR 2026; poster at NDSS 2026)
  • Evading Attention-based Dynamic Malware Detectors — RL-guided adversarial API parameter injection (ACM SAC 2026)
  • Noise-Augmented Transferability — low-query-budget transfer attack on Android malware detectors (IEEE Access 2025)
  • Automated Neural Architecture Search for Deepfake Detection (FPS 2024)

LLM Agents for Security

Building and evaluating LLM-based agent systems that automate security analysis, and engineering guidelines for making such systems reliable.

  • Detecting Phishing on Shared-Domain Hosting — LLM-based contextual mismatch reasoning (ARES 2026)
  • Engineering Guide for Open LLM Agent Systems for Security — SCIS 2026 Paper Award (SCIS 2026)
  • CHASE — LLM agents for dissecting malicious PyPI packages (AIware 2025 @ ASE)
  • BrandSpotter — identifying targeted brand names in phishing with task-optimized lightweight LLMs (IEICE Trans. 2025)
  • LLM-based Privacy-Policy Understandability — automated assessment with LLMs (USEC @ NDSS 2025)

Autonomous Vehicle & Physical AI Security

Offensive and defensive research on autonomous driving and embodied AI: LiDAR perception, traffic-sign and traffic-light recognition, visual SLAM, HD maps, and in-vehicle (CAN) networks.

  • Artistic Adversarial Examples — graffiti-based attacks on traffic sign recognition (ICICS 2026)
  • Cloaking Signals with Signs — traffic-sign-camouflaged patches against traffic light detection (VehicleSec 2026)
  • TEV-IDS — CAN intrusion detection via spatial temporal-entropy-variation fingerprinting (IEEE VTC2026-Spring)
  • Adversarial Fog — exploiting LiDAR point-cloud preprocessing filters, Best Paper Award (ACM AsiaCCS 2025); extended defense study (JIP 2026)
  • Shadow Hack — physical adversarial shadow attack and defense on LiDAR object detection (USENIX Security 2025)
  • vSLAM & Localization Attacks — false-localization on vSLAM drones, NDT vulnerabilities, HD map tampering (VehicleSec 2025)
  • DAP — dynamic adversarial patch attack to feature-based visual odometry (VehicleSec 2025)
  • AVATAR — adversarial vehicle trajectory attack on autonomous-driving planners (ACSW 2025)
  • DeGhost — unmasking phantom intrusions in autonomous recognition systems, Distinguished Paper Award Runner-Up (IEEE EuroS&P 2024)
  • Adversarial Retroreflective Patches — stealthy attack on traffic sign recognition at night (VehicleSec 2024)

Web, Network & Anti-Fraud Security

Measurement and detection at Internet scale: phishing, malicious domains, IDN homographs, web tracking, and email authentication.

  • Large-Scale Analysis of Malware Distribution on YouTube — fake game cheats and cracked software (WACCO 2026 @ IEEE EuroS&P)
  • Melting Pot of Origins — compromising intermediary web services that rehost websites, Distinguished Paper Award (NDSS 2020); journal version (JIP 2024)
  • BIMI — a first look at Brand Indicators for Message Identification (PAM 2023)
  • COVID-19 Domain Names — origin and implications (PAM 2021)
  • Phishing via TLS Certificates — identifying HTTPSified phishing sites by certificate footprints (IEEE EuroS&P Workshops 2020)
  • ShamFinder — automated framework for detecting IDN homographs (ACM IMC 2019) [GitHub]
  • Web Tracking in the Wild — scalable tracking-detection system (IEICE Trans. 2017) and canvas-fingerprinting measurement (RAID 2015)
  • DomainProfiler / DomainChroma — early discovery of abused domains (IEEE/IFIP DSN 2016) and actionable countermeasures (Computers & Security 2018)
  • SFMap — inferring hostnames of encrypted HTTP traffic via domain-name graphs (IFIP TMA 2015)
  • AutoBLG — automated URL blacklist generation (IEEE ISCC 2015)

Mobile & Software Security

Large-scale analysis of mobile app ecosystems, app-behavior consistency, malware analytics, and emerging platforms such as VR.

  • Action Required — mixed-methods study of security practices in GitHub Actions (NDSS 2026)
  • Smart Contract Tooling — evaluation of vulnerability-detection tools (ACM SAC 2026) and random-number generation on smart contracts (ACM DLT 2025)
  • VR App Security & Privacy — investigation via URL string analysis (JIP 2024)
  • Malware Analytics — SmmPack SMM obfuscation (DIMVA 2024); EIGER automated IOC generation (ACSAC 2019); fake-review generation (ESORICS 2018)
  • CLAP — classifying Android PUAs by similarity of DNS queries (IEICE Trans. 2020)
  • Abandoned Internet Resources — threats from resources reused by Android apps (ACM AsiaCCS 2018)
  • Mobile App Vulnerabilities — origins tied to third-party software modules, large-scale measurement (MSR 2017)
  • Third-Party Android Marketplaces — understanding their security management (APWG eCrime 2017)
  • AppRaiser — detecting cloned Android apps at scale (IEICE Trans. 2017)
  • ACODE — app-behavior/description consistency via static analysis and NLP (SOUPS 2015)

Usable Security, Privacy & Human Factors

How people perceive and act on security and privacy, novel authentication mechanisms, and analysis of privacy policies and personal-data ecosystems.

  • Negative Effects of Social Triggers — chains of risky user security/privacy behaviors (SOUPS 2024; JIP 2025)
  • Rhythm Tap — inclusive rhythmic-variation authentication (NDSS 2025 Poster)
  • What Are Cars Collecting? — a study of privacy policies in the automotive industry (VehicleSec 2025)
  • PinchKey — usable behavioral-biometric authentication for VR users (EuroUSEC 2023; JIP 2025)
  • BlinkAuth — recognizing users by their blinks (ACM AsiaCCS 2024)
  • Privacy Policy Analysis — group-company structure and Japanese law revision (WPES 2023)
  • Fraudulent Online-Banking Detection (JIP 2023); online-auction purchase-history privacy (IPSJ 2018)
  • Non-Experts’ Security & Privacy Questions — analysis on a Q&A site (SOUPS 2022; IEICE Trans. 2023)
  • Phishing and Non-Native English Speakers — experiences and behavioral tendencies (JIP 2022); screening-method impact, Best Paper Award (EuroUSEC 2021)
  • GAEN Exposure Notification Privacy — feasibility of linking attacks (PETS 2022)
  • Password Selection across Language Spheres — comparative analysis of linguistic/cultural differences (EuroUSEC 2019; IEICE Trans. 2020)
  • Social-Account De-anonymization — attacker-controllable side channel via user blocking (IEEE EuroS&P 2018)

Offensive & Cyber-Physical Security

Exploring emerging attack surfaces where the physical and digital worlds meet: touchscreens, RF/EM side channels, voice assistants, medical devices, and analog-signal defenses.

Open-Source Code

Selected code released with our papers (see also the full Repos page):

This page highlights representative works; many additional papers, posters, and demos are listed on our achievements pages. Details of ongoing projects will be posted as they become publicly available.