Federated learning is the new frontier, as centralised machine learning becomes outdated

Dr. Arindam Sarkar  
Head of Department of C S & Electronics, Ramakrishna Mission Vidyamandira- Howrah

“Data has become the new oil, the new gold—it's everything. We deal with massive amounts of data, but the challenge lies in managing it. Criminals use land records and other documents to tamper with biometric authentication. In response, we're shifting to Virtual IDs (VIDs), thinking they are more secure. However, the apps we download for conveniences often contain malicious software, putting our data at risk.

We have seen a rise in scams, especially involving loan apps. Cybercriminals use generative AI extensively. They can alter messages, including timestamps, in real-time. To combat this, we need to focus on three key areas: educating people about cyber threats, adopting resilient processes, and ensuring adequate defense mechanisms.

Generative AI and large language models (LLMs) can be used for cyberattacks. Social engineering, phishing, and fake impersonation are becoming more sophisticated with AI. Generative AI acts as a double-edged sword; it can be vulnerable, but it can also be used to defend against attacks. By advancing automated responses and enhancing security measures, we can protect against AI-driven threats.

Generative AI is a subset of deep learning, consisting of two components: a generator and a discriminator. The generator creates new data, while the discriminator evaluates its authenticity. This technology is crucial for fraud analysis. Often, the actual crime data and reported data differ significantly, and generative AI can help fill in these gaps.

Federated learning is the new frontier, replacing centralized machine learning. It allows multiple organizations to collaborate on machine learning models without sharing their data directly, enhancing privacy and security. This shift to federated learning addresses the concerns of companies that are reluctant to share their data, even in cloud environments.”