🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

This attack in Tbilisi shows how vulnerable public infrastructure is to cyberattacks, even seemingly benign systems like payment kiosks. #IIoT #cybersecurity #criticalinfrastructure

Excited to see our #ICLR2025 work featured! We introduce AoT+, which leverages memoization to boost planning reliability. I'm especially excited about continue pushing towards token reduction & real-time compute for a greener AI! 🌿Read more here: devdiscourse.com/article/techno… #AI

Cyberattacks on utilities are surging (via Reuters: shorturl.at/RmvC4)! Our DOE-funded research uses AI (in-context learning without ground truth labels) to detect zero-day attacks in power substations with >85% accuracy. Read more: arxiv.org/abs/2501.16453 Partners:

Trustworthy LLMs are crucial! Our new review paper explores Uncertainty Estimation in LLMs, covering sources, methods, and challenges. Essential reading for anyone working with LLMs. Check it out: hal.science/hal-04973361 #LLM #Uncertainty #SafeAI The Sanghani Center at Virginia Tech

Honoring Prof. Chen-Ching Liu’s incredible legacy. His mentorship and collaborative spirit have left a lasting impact on our community and research. Grateful for the opportunities to work together on cybersecurity for power systems.

Energized by the 2025 PEC Conference — a forum where real-world challenges meet technical innovation. From AI to grid resilience, it’s clear that the future of power systems needs interdisciplinary collaborations. Grateful to see our students taking initiative and connecting with

Excited to share that our team just won CyberFarm 2025! We developed a real-time, LLM-based intrusion detection system to secure agricultural water systems—protecting the infrastructure behind our food supply. #AI #LLMs #Cybersecurity #AgTech The Sanghani Center at Virginia Tech Virginia Tech ECE

🧵New survey: Bridging Distribution Shift and AI Safety Distribution shift and AI safety have long been studied in parallel. But how can their insights formally inform each other? We present the first comprehensive, mathematically grounded, and one-to-one aligned treatment. 1/6

For our #ICML2025 position paper, we argue AI safety should shift from brittle fortresses to adaptive immune systems. This "antifragile" approach builds systems that learn and strengthen from real-world stress. 📄 Paper: jinming.tech/papers/Antifra… 📊 Slides:

Thrilled to host Tangyu Ma Tengyu Ma (Stanford) at the AI Agent Frontier Seminar (sites.google.com/view/ai-agent-…) this Friday! He'll present "STP: Self-play LLM Theorem Provers with Iterative Conjecturing and Proving." A fantastic talk on the future of AI agent and theorem proving. 🗓️