CNI CONNECT

Monthly Newsletter from CNI

Issue - October 2025

CodeIT 2025: Workshop on Codes, Sequences and Information Theory

The Centre for Networked Intelligence (CNI) is organizing CodeIT 2025, a two-day workshop to be held from 11–12 October, featuring invited talks by distinguished national and international experts, covering recent advances in coding theory, sequence design, and information theory.

The workshop, held in honor of Prof. P. Vijay Kumar, aims to foster scholarly exchange and engage students, researchers, and practitioners working at the intersection of communication and information sciences.

Further details are available on the CNI workshop webpage.

Upcoming Talk

“On Pricing for Routing and Flow-Control in Payment Channel Networks” by Prof. Bruce Hajek (University of Illinois Urbana-Champaign) on 14 October 2025

Stay tuned for more talks:

on 21 October 2025 by Prof. Saurabh Jain from IIT Bombay

on 28 October 2025 by Prof. Abhinav Kumar from IIT Hyderabad

CNI recently hosted the Generative AI Hackathon 2025, bringing together talented minds to explore innovative applications of generative AI for India-centric challenges.

With 3,292 registrations from 100 institutes across India, the event witnessed exceptional participation from students across disciplines.

The results were announced on 26 September 2025, recognizing the most creative solutions.

Read the full results and highlights here.

CNI is inviting applications for the following positions:

l Research Associate: For experienced professionals and students planning to join after graduation.

l Research Intern: For students who wish to collaborate with CNI while pursuing their studies.

If you are passionate about research in networks, AI, and systems, and wish to work with an interdisciplinary team at IISc, we encourage you to apply. Click here to know more.

Intelligent Resource Management: from LLM Training to Web Serving

Dr. Rohan Gandhi (Microsoft Research India) discussed efficient use of computing resources through two projects; Atlas and KnapsackLB. Atlas improves GPU utilization for faster LLM training, while KnapsackLB balances web traffic, cutting latency and enhancing system reliability and cost-effectiveness. The talk highlighted how intelligent resource management can make AI systems faster, more reliable, and more cost-effective.

Signaling Techniques for Green Communications

Prof. Ranjan K. Mallik (IIT Delhi) spoke on energy-efficient communication systems using low-power receivers and noncoherent signaling. He compared receiver designs and showed how smart signal processing methods can make wireless communication greener, simpler, and more practical for future networks.

When Privacy meets Partial Information: Privacy-Utility Trade-offs in Bandits

Prof. Debabrata Basu (INRIA France) explained how algorithms can protect sensitive data while learning efficiently. He showed that adding Differential Privacy to bandit models maintains performance, offering new insights into balancing learning efficiency and data protection in modern systems.

Detecting AI Slop in Research and Beyond

Prof. Danish Pruthi (IISc Bangalore) discussed detecting AI-generated and plagiarized research content. He explained his team’s methods, such as watermarking, to identify AI text. The talk sparked conversations on ethics, reliability, and responsible AI use in academic writing.

Decoupling Theorem in Quantum Information Theory: Application and Derandomization

Prof. Aditya Nema (IIT Gandhinagar) explained the decoupling theorem’s role in simplifying quantum states. He presented a new approach using pseudorandom unitaries, showing how derandomization improves the efficiency and scalability of quantum information processing.

Dynamic Content Caching and Delivery using Restless Multi-armed Bandits

Ankita Koley (DESE, IISc) presented caching methods that improve online content delivery. Using a restless multi-armed bandit model and a “wait” option, she proposed near-optimal, cost-effective strategies to balance freshness, cost, and performance in data networks.