ACM SIGSAC Featured My Work on LinkedIn
I was featured by the ACM Special Interest Group on Security, Audit, and Control (SIGSAC) on LinkedIn.
Data-Driven Security & Privacy Lab (DSPLab)
We build trustworthy AI systems that are robust, privacy-preserving, and auditable. Our work sits at the intersection of AI security, adversarial machine learning, provenance, and deployable safety.
Latest Lab News
DeepLeak Accepted at IEEE SaTML
Our latest paper "DeepLeak: Privacy Enhancing Hardening of Model Explanations Against Membership Leakage" is accepted to the IEEE Conference on Secure and Trustworthy Machine …
Keynote at IEEE SaTC 2026
I will be giving a keynote at the IEEE Conference on Secure and Trustworthy CyberInfrastructure for IoT and Microelectronics (SaTC 2026) in Houston, TX.
Research Areas
Trustworthy AI Across the Lifecycle
Our work spans AI security, privacy, safety, explainability, and ethical norms.
Active
AI Security
Hardening AI against backdoors, data poisoning, evasion, and model-stealing threats.
Backdoor Detection
Poisoning Defense
Adversarial Robustness
Active
AI Privacy
Protecting training data and model internals from inference attacks and sensitive leakage.
Membership Inference
Leakage Measurement
Privacy-Preserving ML
Active
AI Safety
Ensuring models behave safely and predictably in high-stakes and adversarial settings.
Unsafe Behavior Detection
Assurance Mechanisms
Reliable Deployment
Active
Explainability and Accountability
Building interpretable and auditable AI where decisions can be traced and validated.
Explanation Robustness
Decision Traceability
Forensic Analysis
Active
Ethical and Responsible AI
Embedding responsible deployment norms across the AI lifecycle in critical systems.
Responsible AI Practice
Ethical Norms
Accountability-by-Design
Provenance-Centric AI Security and Safety
DSPLab research advances a new paradigm for securing and validating artificial intelligence systems: Provenance-Centric AI Security and Safety. As AI systems increasingly influence high-stakes domains such as cybersecurity, finance, healthcare, and autonomous systems, the core challenge is no longer only improving accuracy but ensuring that AI systems remain robust, safe, trustworthy, and accountable throughout their lifecycle. We argue that the key to achieving this lies in provenance: understanding and tracing the origins, lineage of transformations, and influence pathways that shape a model’s behavior. Traditional AI evaluation relies largely on black-box testing, observing outputs without visibility into the internal processes that produced them, which leaves critical blind spots against threats such as data poisoning, backdoor attacks, adversarial manipulation, and unsafe or unintended model behaviors. Our work introduces fine-grained observability into the AI pipeline by tracking the lifecycle history of data, training dynamics, parameter updates, and inference-time information flows. Through this provenance-centric lens, we develop the theoretical foundations, algorithms, and systems that make robustness and safety measurable, explainable, and auditable, enabling capabilities such as attack detection, forensic analysis, accountability, and automated model repair. Our vision is to establish provenance as a foundational layer for AI security and AI safety, transforming AI from opaque systems into observable and auditable infrastructures where model decisions can be traced, inspected, and verified, enabling responsible deployment of advanced AI in critical societal systems.
Flagship Projects
Open-source systems and reproducible artifacts from recent lab work.
DeepLeak
Privacy hardening for explanation methods against membership inference leakage.
DeepProv
Inference provenance graph analysis for behavioral diagnosis and targeted DNN repair.
PoisonSpot
Fine-grained training provenance tracking to detect clean-label backdoor poisoning.
Featured Publications
Recent papers from DSPLab at flagship security and trustworthy AI venues.
DeepLeak: Privacy Enhancing Hardening of Model Explanations Against Membership Leakage
Machine learning (ML) explainability is central to algorithmic transparency in high-stakes settings such as predictive diagnostics and loan approval. Yet these same domains demand …
DeepProv: Behavioral Characterization and Repair of Neural Networks via Inference Provenance Graph Analysis
Deep neural networks (DNNs) are increasingly being deployed in high-stakes applications, from self-driving cars to biometric authentication. However, their unpredictable and …
PoisonSpot: Precise Spotting of Clean-Label Backdoors via Fine-Grained Training Provenance Tracking
Relying on untrusted data exposes machine learning models to backdoor attacks, where adversaries poison training data to embed hidden behaviors. Existing defenses struggle against …
Meet the Team
Researchers Building Trustworthy AI
We are a collaborative group of researchers working across AI security, privacy, and dependable deployment.
Principal Investigator
Birhanu Eshete
Principal Investigator
University of Michigan-Dearborn
Associate Professor of Computer Science at the University of Michigan-Dearborn and Director of the Data-Driven Security & Privacy Lab (DSPLab).
AI Trustworthiness
AI Security
Privacy-Preserving ML
PhD Researchers
MS Researchers
Alumni
Abe Amich
R&D ML Engineer for Cybersecurity, Sandbox AQ
Ph.D., 2019-2024, University of Michigan-Dearborn.
Christine Carlton
Network Monitoring and Observability Manager, Ford Motor Company
M.Sc., 2023, University of Michigan-Dearborn.
Ata Kaboudi
Software Engineer, CBRE Investment Management
M.Sc., 2023, University of Michigan-Dearborn.
Jon-Nicklaus Jackson
IT Security and Compliance Analyst, Bosch USA
M.Sc., 2023, University of Michigan-Dearborn.
Chevy Pawlik
IT Security Analyst, Auto-Owners Insurance
B.Sc., 2022, University of Michigan-Dearborn.
Youssef Aydi
M.S. Student, University of Michigan-Dearborn
B.Sc., 2024, University of Michigan-Dearborn.
Zeineb Moalla
M.S. Student, University of Michigan-Dearborn
B.Sc., 2022, University of Michigan-Dearborn.
Join DSPLab
Open positions are not currently available. We welcome inquiries from students interested in future opportunities in trustworthy AI, AI security, and privacy.
Start the Conversation
We are always open to discussing new projects, opportunities, or just having a chat.
Visit Us
Data-Driven Security & Privacy Lab (DSPLab)
Department of Computer and Information Science
University of Michigan-Dearborn
Dearborn, MI, USA
Office Hours
By appointment
Connect
Prospective Students
Open positions are not currently available. You can still share your CV and interests if you would like to be considered for future opportunities.
Email Prof. Eshete