Chashi Mahiul Islam
Papers
1
Total Citations
5
H-Index
1
About
Chashi Mahiul Islam is a rising researcher at the intersection of artificial intelligence, cybersecurity, and multi-modal systems. His primary research areas include vision-language navigation (VLN), adversarial machine learning, and the security vulnerabilities of large language models and vision-language transformers. Islam’s most notable contribution is his pioneering work on exposing representation vulnerabilities in VLN systems, as demonstrated in his highly cited 2024 paper, "Malicious Path Manipulations via Exploitation of Representation Vulnerabilities of Vision-Language Navigation Systems." This work reveals how adversaries can exploit the internal representations of multi-modal models to manipulate navigation paths, even when the systems leverage state-of-the-art large language models for command understanding and zero-shot recognition. By identifying these critical security gaps, Islam has laid essential groundwork for developing more robust and trustworthy AI navigation agents. His research has already garnered significant attention, with his top paper accumulating 5 citations in a short time, signaling its impact on both the AI safety and robotics communities. Islam’s work is particularly relevant as VLN systems move toward real-world deployment, making his contributions vital for ensuring their reliability against malicious attacks.
Research Focus
Key Achievements
Top Papers
- 1