Howard University Secures $700,000 NSF Grant to Study Robust Machine Learning

Howard University has secured a $700,000 research award from the US National Science Foundation to fund ongoing investigations into resilient machine learning systems.

The project is led by principal investigator Danda B. Rawat, a professor of computer science in the College of Engineering and Architecture. Rawat also serves as the associate dean for research and graduate studies at the college, alongside his role as director of the university's Data Science and Cybersecurity Centre.

The grant supports foundational research aimed at resolving critical vulnerabilities across modern machine learning architectures. While contemporary artificial intelligence models have demonstrated high empirical performance, their core algorithms often remain fragile when deployed in unconstrained operational settings. Rawat's research programme focuses on developing computational frameworks that remain reliable, robust, and secure against adversarial manipulation, noise, and data corruption.

Securing machine learning systems requires mitigating structural weaknesses at the algorithmic level. Adversarial inputs, model poisoning, and domain shifts present distinct failure modes for complex systems deployed across critical infrastructure. Rawat's project seeks to address these challenges by examining the underlying mathematical principles that govern statistical learning, intending to establish rigorous guarantees of model reliability and trustworthiness.

The funding also directly supports academic training for both graduate and undergraduate students at Howard University. By integrating advanced machine learning research into student degree pathways, the initiative seeks to expand technical capacity and prepare researchers from historically underrepresented backgrounds for specialised careers in computer science and cyber-defence.

Federal investments of this nature reflect sustained institutional interest in academic AI research outside the commercial sector. As private industry concentrates computational capital on scaling existing paradigms, university laboratories continue to provide theoretical foundations for system verification, robustness, and mathematical safety.

As deployment environments grow increasingly hostile to unverified neural architectures, can foundational academic research keep pace with the systemic vulnerabilities emerging from frontier models?