IIT Madras’ Centre for Responsible AI and Ericsson partner for joint research in Responsible AI

CHENNAI, 26 September 2023: Indian Institute of Technology Madras’ (IIT Madras) Centre for Responsible AI (CeRAI) right now introduced that it’s partnering Ericsson for joint research in the realm of Responsible AI. To commemorate the event, a Symposium on Responsible AI for Networks of the Future was organized the place leaders from Ericsson Research and IIT Madras participated to debate the developments and developments in the sector of Responsible AI.

During the occasion held on the IIT Madras campus right now, Ericsson signed an settlement to partner with CeRAI as a ‘Platinum Consortium Member’ for 5 years. Under this MoU, Ericsson Research will assist and take part in all research actions at CeRAI.

The Centre for Responsible AI is an interdisciplinary research centre that envisions turning into a premier research centre for each elementary and utilized research in Responsible AI with instant impression in deploying AI methods in the Indian ecosystem.

AI Research is of excessive significance to Ericsson because the 6G networks could be autonomously pushed by AI algorithms.

Addressing the symposium, Chief Guest, Prof. Manu Santhanam, Dean (Industrial Consultancy and Sponsored Research), IIT Madras, stated,“Research on AI will produce the instruments for working tomorrow’s companies. IIT Madras strongly believes in impactful translational work in collaboration with the trade, and we’re very completely satisfied to collaborate with Ericsson to do innovative R&D in this topic.”

Speaking on the event,Dr. Magnus Frodigh, Global Head of Ericsson Research, stated,“6G and future networks purpose to seamlessly mix the bodily and digital worlds, enabling immersive AR/VR experiences. While AI-controlled sensors join people and machines, accountable AI practices are important to make sure belief, equity, and privateness compliance. Our focus is on growing cutting-edge strategies to boost belief and explainability in AI algorithms for the general public good. Our partnership with CERAI at IIT Madras is aligned with Indian Government’s imaginative and prescient for the Bharat 6G program.”.

A panel dialogue on ‘Responsible AI for Networks of the longer term’ was organised to commemorate the partnership through the symposium and a few of the present research actions being carried out on the Center for Responsible AI had been showcased.

Elaborating on the partnership between CeRAI and Ericsson, Prof. B. Ravindran, Faculty Head, CeRAI, IIT Madras, andRobert Bosch Centre for Data Science and AI(RBCDSAI), IIT Madras, stated,“Networks of the longer term will allow simpler entry to excessive performing AI methods. It is crucial that we embed accountable AI ideas from the very starting in such methods. Ericsson, being a pacesetter in future networks is a perfect partner for CeRAI to drive the research and for facilitating adoption of accountable design of AI methods.”

Speaking concerning the work that may be taken up beneath this collaboration, Prof. B. Ravindran added,“With the arrival of 5G and 6G networks, many crucial functions are prone to be deployed on units comparable to cell phones. This requires new research to make sure that AI fashions and their predictions are explainable and to offer efficiency ensures acceptable to the functions they’re deployed in.”

The Speakers and Panellists of the Symposium embrace Prof. R. David Koilpillai, Qualcomm Institute Chair Professor, IIT Madras, Dr. Harish Guruprasad, Core Member, CeRAI, IIT Madras, Dr. Arun Rajkumar, Core member – CeRAI, Dr. Jorgen Gustafsson, Head of AI , Ericsson Research, Dr. Catrin Granbom, Head of Cloud Systems and Platforms, Ericsson Research, Kaushik Dey, Research Leader -AI/ML, Ericsson Research – India

Some of the important thing initiatives offered throughout this Symposium embrace:

• The mission on large-language fashions (LLMs)in healthcare, which focuses on detecting biases proven by the fashions, scoring strategies for real-world applicability of a mannequin, and lowering biases in Large Language Models (LLMs). Custom-scoring strategies are being designed primarily based on Risk Management Framework (RMF) put forth by National Institute of Standards and Technology (NIST), the U.S. federal company for advancing measurement science and requirements.
• The mission on participatory AI addresses the black-box nature of AI at numerous phases, together with pre-development, design, growth and coaching, deployment, post-deployment and audit. Taking inspiration from domains comparable to city planning and forest rights, the mission research governance mechanisms that allow stakeholders to offer constructive inputs for higher customization of AI, enhance accuracy and reliability, increase objections over potential destructive impacts.
• Generative AI fashions primarily based on consideration mechanismshave not too long ago gained vital curiosity for their distinctive efficiency in numerous duties comparable to machine translation, picture summarization, textual content technology, and healthcare, however they’re complicated and tough for customers to interpret. The mission on interpretability of attention-based fashions explores the circumstances beneath which these fashions are correct however fail to be interpretable, algorithms which may enhance the interpretability of such fashions, and understanding which patterns in the information these fashions are inclined to study.
• Multi Agent Reinforcement Learning for trade-off and battle decisionin intent primarily based networks: Intent-based administration is gaining traction in telecom networks attributable to strict efficiency calls for. Existing approaches typically use conventional strategies, treating every closed loop independently and missing scalability. This mission research a Multi-agent Reinforcement Learning (MARL) technique to deal with complicated coordination and encouraging loops to cooperate mechanically when intents battle. Current efforts discover generalization skills of the mannequin by leveraging explainability and causality for joint actions of brokers.