Prof. Sudip Misra, IEEE Fellow
IIT Kharagpur
Email: sudipm@iitkgp.ac.in

Abstract: The introduction of IoT in healthcare has led to many advancements and added new dimensions to the traditional healthcare systems. The interactions between patients and doctors have moved beyond physical visits with the integration of IoT in the healthcare domain. Digital storage of medical data, remote health monitoring, and historical data analysis are some of the significant developments that have been accomplished in the area of healthcare. Wearable devices have played a vital role and have proved to be a game-changer in embedding IoT in healthcare. In the present scenario, wearables have witnessed growing popularity and have become a part of people's lifestyles. Wearables have enabled continuous monitoring of physiological parameters, both locally and remotely. Monitoring the elderly, children, and patients with chronic illnesses have especially benefitted with the availability of wearables. In addition to this, more complex systems for the diagnosis of diseases provide better insights into the patient's condition, which helps in early detection, faster response time, and targeted treatment. IoT-enabled healthcare connects patients from less accessible rural areas to remote healthcare professionals. In this keynote, we will discuss the state-of-the-art in the e-Health domain and its evolution with the introduction of wearables and diagnostics systems.

Prof. Katarzyna Wasielewska-Michniewska
Systems Research Institute, Polish Academy of Sciencesul.
Newelska 6, 01-447 Warsaw
Email: katarzyna.Wasielewska@ibspan.waw.pl

Abstract: Federated learning (FL) was proposed to train models in distributed environments allowing to mitigate some problems related to data privacy and usage of local resources. Until now, a lot of research has been devoted to the foundamental issues, such as adaptation of machine learning algorithms to the FL constraints, data privacy protection, or dealing with effects of unbalanced data distribution. However, not much attention is paid to the requirements of actual FL deployments coming from real-life use cases. In this context, in the presentation the motivation, major assumptions and challenges for applying FL to IoT ecosystems will be outlined. Moreover, an architecture that enables building flexible and adaptable FL solutions will be introduced. The practical use case will be discussed, where FL is to be actually deployed within an Internet of Things ecosystem. Hence, different issues that need to be considered in such situation will be identified.

Prof. Paweł Szmeja
Systems Research Institute, Polish Academy of Sciences
Email: Pawel.Szmeja@ibspan.waw.pl

Abstract: Knowledge graphs are one of the technologies that is the backbone of many modern systems, including search engines, databases, and others. Underpinned by mathematical logic, they provide a way to store highly connected data and reason over it, discovering information that is often not explicit, but highly useful. The talk will introduce the knowledge graphs and their applications, focusing on the problem of similarity. The everyday problem of similarity takes on a new meaning, when one considers, that there are hundreds of similarity measures, defined for different kinds of data representations and aimed and solving highly non-trivial problems in the domain of knowledge management. The talk outlines these measures, as well as different ways to represent knowledge in a graph, to culminate in brief answers to critical questions: how and why to calculate similarity in a knowledge graph.

Prof. Prasanta K. Jana, IEEE Senior Member
IIT (ISM), Dhanbad
Email: prasantajana@iitism.ac.in

Abstract: Internet of Things (IoT) has become an inevitable technology in which a huge number of devices are connected through Internet for providing instant services to the end users in various application areas such as smart cities, autonomous vehicles, smart healthcare, smart agriculture, smart transportation and so on. In order to obtain insights into data generated from the IoT devices, machine learning (ML) techniques such as decision tree, support vector machine, deep learning etc. have been widely used in such applications. Traditionally, data is sent to a cloud server to train data models. However, to achieve an efficient ML model under this conventional centralized approach raises privacy issue, particularly when the clients need to discourse their personal or sensitive data to the cloud server. Federated Learning (FL) has become an efficient technique to address this issue in which model training can be performed without sharing data with the cloud data center so that data privacy can be preserved at the client levels. However, client selection, straggler effects, communication cost, data aggregation etc. are some of the challenging issues that must properly addressed while producing an ML model in the FL platform. In this talk, I will discuss with all these issues of the FL with its current trend and future research works. The presentation will begin with the necessary background and motivation of this paradigm. Next, I will describe about various proposed FL architectures with their merits and demerits. This will be followed by some exiting FL schemes addressing the issues of communication cost, straggler effects and other matters of FL in the IoT platform. The session will be ended by future research directions of FL.

Prof. Amiya Kumar Rath
Biju Patnaik University of Technology, Rourkela & Veer Surendra Sai University of Technology, Burla
Email: vc@bput.ac.in, amiyakumarrath@gmail.com

Abstract: The influence of the Internet of Things (IoT) in the context of the Smart City mission has been transformative and far-reaching. The integration of IoT technologies within urban environments has led to the creation of intelligent and interconnected systems that enhance various aspects of city living. IoT enables real-time data collection and analysis from diverse sources such as sensors, devices, and infrastructure, enabling city planners to make informed decisions for efficient resource management, traffic optimization, waste management, energy conservation, and more. This synergy between IoT and the Smart City mission fosters sustainability, innovation, and improved quality of life for residents by establishing a dynamic ecosystem that responds to the evolving needs of urban areas.The influence of the Internet of Things (IoT) on various domains like smart water management, smart grid systems, smart agriculture, and smart parking has redefined the landscape of urban living. In the realm of smart water management, IoT-enabled sensors and data analytics empower cities to monitor water quality, consumption patterns, and leakage detection in real time, leading to efficient resource allocation and conservation. Smart grid systems leverage IoT to optimize energy distribution, monitor power usage, and enable demand-response mechanisms, thereby enhancing energy efficiency and sustainability. In the context of smart agriculture, IoT technologies facilitate precision farming by enabling remote monitoring of soil conditions, crop health, and irrigation, resulting in higher yields and minimized environmental impact. Additionally, IoT-driven smart parking solutions offer real-time information about parking availability, reducing traffic congestion and carbon emissions. Overall, the integration of IoT into these sectors exemplifies the potential for data-driven innovation to revolutionize urban infrastructure and enhance the quality of life for residents.

Prof. Tanmoy Chakraborty
Indian Institute of Technology, Delhi
Email: tanchak@iitd.ac.in

Abstract: Online social media platforms are popular mediums for disseminating and consuming information. Unfortunately, due to the decentralized generation and propagation of content, they also come with limited liability for harmful posts and collusive persuasion. It is often the experience that online users are subjected to a barrage of harmful posts (fake news, hate speech, offensive posts, etc.) within a short span of time. In this talk, I shall present one of our megaprojects, "Project Robinhood", which we envision as an all-in-one solution to curb the spread of harmful content on social media. In particular, I will present a series of our recent studies on combating fake news, hate speech and harmful content. I will discuss how proactive measures can help stop the spread of hate speech even before it is written! The talk will end with a glimpse of other ongoing research at the LCS2@IITD lab.

Prof. Siba K Udgata
WiSECom Lab
School of Computer and Information Sciences
University of Hyderabad, Hyderabad, India, 500046
Email: udgata@uohyd.ac.in

Abstract: Device-free Wi-Fi sensing has gained much attention due to its simplicity, low cost, and it requires no additional hardware. The main advantages of Wi-Fi sensing are that it is unobtrusive, can operate through walls, works without lighting, is ubiquitous, and does not require users to carry additional wearable devices. CISCO estimates there will be 543 million Wi-Fi hot spots worldwide by the end of 2022, making Wi-Fi signal availability almost omnipresent. The traditional methods of video and sensor-based systems suffer from acceptability, availability, affordability, and privacy concerns, which need to be improved. The received Wi-Fi signal characteristics change with a change in the dielectric constant of the medium and other reflections and scattering. These changes in the Wi-Fi signal patterns can be exploited to detect various events and environmental conditions in the wireless zone, identify the materials in the wall, and detect faults. Recently Wi-Fi sensing techniques are also being used to measure physiological parameters like heartbeat and breathing rate monitoring, among other things. This talk attempts to demonstrate a device-free general-purpose Wi-Fi sensing system to track events and recognize activities even through the wall and other materials using the Channel State Information (CSI) values extracted from the received Wi-Fi signals at the receiver end. The received signal characteristics change with the presence of human beings, and their activities affect signal propagation, resulting from reflection and scattering. The activities can be recognized by analyzing the CSI values corresponding to different sub-carriers of the received signal. CSI values contain fine-grain information such as amplitude and phase to achieve better sensing accuracy with a unique pattern that can be observed corresponding to each activity and material. We will present our experience of developing the transmitter and receiver hardware modules together with the necessary software for capturing the CSI from Wi-Fi signals and conduct of multiple experiments using a low-cost ESP-32 module and development machine learning models for human presence detection, activity detection, material classification, intrusion detection, an ambient condition in indoor environments among a few others. The results show the effectiveness of the proposed method for different applications of technological and societal importance.

Prof. Saroj Kumar Meher
Indian Statistical Institute, Bangalore
Email: saroj.meher@isibang.ac.in

Abstract: Domain adaptation is an essential technique in the field of pattern classification that intends to address the issues of deploying machine learning models in real-world scenarios where the training data and testing data come from different domains. This technique was developed to address the challenges of deploying machine learning models in real-world scenarios. When discussing pattern classification, the term "domain" refers to a unique data distribution due to many traits, statistical properties, or underlying structures. The primary problem occurs when a classifier trained on a labeled dataset originating from a source domain does not perform effectively when applied to a target domain due to a domain shift. The domain adaptation approaches aim to bridge this gap by effectively transferring knowledge from the source domain to the target domain by learning representations or mappings that align the source and target domains. Domain adaptation is a vital area of research to advance the practical applicability of pattern classification models in various and constantly changing real-world environments. Techniques like adversarial domain adaptation, instance re-weighting, and domain-invariant feature learning have shown promise in improving how well classifiers work across different domains and how well they can generalize. As a result, domain adaptation has become an essential area of research.