Conference Schedule

ConfAI 2026 | October 30 to November 01, 2026 | Plaksha University, Mohali

View Brochure
Day 1: Friday, October 30, 2026
07:00 AM – 08:30 AM
Breakfast
Dining Hall (First Floor)
08:30 AM – 09:30 AM
Arrival / Registration & Refreshments
A1 - MR 1005
09:30 AM – 10:00 AM
Welcome Address & ConfAI Vision
Auditorium
10:00 AM – 10:50 AM
Plenary Keynote | Dr. Santanu Chaudhury
TalkCausal and Explainable AI: Challenge of Solving Real Life Problems
View Abstract
Integrating causality into machine learning provides a principled foundation for achieving robustness under distribution shifts, hallucinations and provides a framework for explainable AI. Causality ensures stability across diverse operational conditions and allows models to learn to respond to interventions and shifting environments. In our talk, we shall show how domain based features can be exploited to ensure robust and explainable AI performance in different domains. We shall be considering two distinct contrasting application domains – medical imaging and campaign designs for commercial interventions.
Auditorium
10:55 AM – 11:05 AM
Tea Break
Open Area Infront of Classroom A3-007
11:05 AM – 12:45 PM
Track Presentations (Parallel Sessions)
Track Presentations: AI in Healthcare
11:15 AM – 12:45 PM: Track Oral Presentations & Flash Talk
View Presentations & Flash Talks
Oral Presentations (15 min each: 12 min talk + 3 min Q&A)
B H V S P Subrahmanyam
Generate, Verify and Select: Evolutionary Fitness Shaping for Safe Clinical Text Generation
Aarju Dixit
FedBio-AD: Federated learning enabled digital biomarker framework for early Alzheimer’s disease detection
Antara Roy Choudhury
Towards Scalable, Explainable Early Detection of Autism Spectrum Disorder through Multimodal Behavior Modeling in Triadic Play
Flash Talk (2 min, 1 slide)
Archit Bhatia
Room: A3 - 007
Track Presentations: AI and Logic: Relations and Applications
11:05 AM – 11:35 AM: Track Keynote: Prof. Akshay S
11:35 AM – 12:45 PM: Track Oral Presentations & Flash Talk
Track Keynote TalkLogic and AI: Old Connections, New Frontiers
View Abstract
Logic and artificial intelligence have shared a rich relationship since the early days of AI. Logical methods have long helped build intelligent systems through knowledge representation, planning, and automated reasoning, while advances in AI are now helping solve hard problems in logic itself, from SAT/SMT solving to theorem proving. The rise of learned systems has brought a new emphasis to this relationship: using formal and symbolic reasoning to understand and analyse the behaviour of learned models.

In this talk, we will explore these evolving connections and illustrate how logical and constraint-based methods enable verification, explainability, and quantification of AI models, using our recent work on decision tree ensembles as an example. We will also take a broader perspective on where this relationship is headed, touching on recent efforts to use logic to characterise the expressive power and limitations of modern learning architectures.
View Presentations & Flash Talks
Oral Presentations (15 min each: 12 min talk + 3 min Q&A)
Ajay M Reje
AI Futures beyond Technological Determinism: Pratītyasamutpāda and the Relationality of Human-Machine Agency
Dr. Rajesh Kumar
Event-centric semantic alignment of vulnerabilities with adversarial attack techniques
Abhishek Gupta
Adapting to Evolving Graphs: A Scalable Framework for Dynamic Coarsening
Flash Talk (2 min, 1 slide)
Jaya Prakash Arjarapu
Room: A3 - 009
AI In Healthcare: A3 - 007


AI and Logic: A3 - 009
12:55 PM – 01:10 PM
Group Photograph
Outside Library Grand Staircase
01:10 PM – 02:00 PM
Lunch
Dining Hall (First Floor)
02:10 PM – 03:00 PM
Plenary Keynote | Dr. Parag Singla
TalkNeuro-Symbolic AI: Enhancing Capabilities of Generative Models and Challenges Ahead.
View Abstract
While modern day deep-networks have advanced the state of AI significantly, with applications spanning almost every discipline, some fundamental challenges remain. Many of these pertain to the nature of tasks that involve symbolic reasoning, where SOTA models still seem to falter. In this talk, I will present a broad overview of some of the work in our group on applying neuro-symbolic techniques to enhance the capabilities of generative models (1) We look at the problem of text-conditioned image generation with fine details. We present a novel technique that formulates generation as a sequence of generate, plan and edit steps, allowing the model to correct the original generation iteratively in an object-centric manner by generating a stepwise edit plan. (2) We look at the problem of representing grounded spatial-concepts using simpler building blocks, e.g., a staircase is a sequence of towers of increasing height and explore how these can be learned by a robotic agent from a few human demonstrations, in the form of executable (Python) programs. Our approach combines MCTS for searching in the plan space, with LLMs to create the initial task sketch from natural language, and to aid generalisation of learned programs.
Auditorium
03:10 PM – 04:50 PM
Track Presentations (Parallel Sessions)
Track Presentations: AI In the Economy
03:10 PM – 03:40 PM: Track Keynote
03:40 PM – 04:50 PM: Track Oral Presentations & Flash Talks
View Presentations & Flash Talks
Oral Presentations (15 min each: 12 min talk + 3 min Q&A)
Zarka Bashir
FC2L: Federated Continual Graph Contrastive Learning for Evolving Cross-Bank Money Laundering Detection
Praveen Parthasarathy
AI career and financial guidance platform designed to close the advisory gap facing Teens
Flash Talk (2 min, 1 slide)
Daksh Mehta
Room: A3 - 007
Track Presentations: AI in Sensing
03:10 PM – 03:40 PM: Track Keynote: Dr. Mohan Kumar
03:40 PM – 04:50 PM: Track Oral Presentations
Track Keynote TalkAI in Sensing: From Raw Signals to Personal Insight in Wearables
View Abstract
Wearables such as smartwatches, rings, patches and earbuds continuously capture physiological and motion signals, including heart rate, ECG, PPG, skin temperature, movement and sleep. Turning this noisy, motion-corrupted data into reliable health insight is where AI makes the difference. This talk explores how machine learning improves signal quality, enables early detection of conditions such as arrhythmia and sleep disorders, and personalizes models to each user. It also examines on-device AI, which delivers real-time, low-power, privacy-preserving inference. Key challenges are discussed, including limited labeled data, variability across users and skin tones, battery constraints, and clinical validation. The talk closes with an outlook on foundation models for biosignals and the path from consumer wellness tracking to trusted medical-grade monitoring.
View Presentations
Oral Presentations (15 min each: 12 min talk + 3 min Q&A)
Dr. Debajyoti Biswas
Exploring Machine Learning and Large Language Models in Microelectronics
Devesh Bajaj
Detecting Asian Elephant Vocalizations Using CNN-LSTM architecture from Acoustic Recordings
Room: A3 - 009
AI In the Economy: A3 - 007


AI in Sensing: A3 - 009
04:50 PM – 05:40 PM
High Tea Break / Poster Presentation
Tracks: Track 1: AI in Healthcare  |  Track 2: AI and Logic  |  Track 3: AI in the Economy  |  Track 4: AI in Sensors
A3 First Floor Porch Area
05:40 PM – 06:00 PM
Campus Tour
Plaksha Campus
06:00 PM – 07:15 PM
Break
Campus Area
07:15 PM Onwards
Networking Dinner
Dining Hall ( First Floor )
Day 2: Saturday, October 31, 2026
07:00 AM – 08:30 AM
Breakfast
Dining Hall (First Floor)
08:30 AM – 09:00 AM
Refreshments
A1 - MR 1005
09:00 AM – 09:10 AM
Day 2 Opening Remarks
Auditorium
09:10 AM – 10:00 AM
Plenary Keynote | Prof. Sanghamitra Bandyopadhyay
TalkComputational Methods for Life Sciences
View Abstract
The application of computational methods to biological data has transformed modern biological research, enabling researchers to identify complex patterns, uncover trends, infer relationships, and detect anomalies that may not be readily apparent through conventional approaches. This has accelerated the pace of biological discovery and provided deeper insights into the molecular and cellular processes underlying life, health, and disease.

In this talk, I will begin with a brief overview of molecular biology, including the central dogma that describes the flow of genetic information from DNA to RNA and proteins. The diverse types of biological and healthcare will then be introduced. The talk will subsequently focus on selected computational methods developed in our group at the Indian Statistical Institute for biological data analysis and healthcare applications. These will include classification-based approaches for predicting microRNA targets, clustering and computational analysis of single-cell RNA sequencing data, and multiobjective optimization approaches for drug design that seek to balance multiple competing objectives. An ongoing work on the analysis and prediction of adverse events arising from polypharmacy, an increasingly important problem in clinical practice, will also be discussed.

The talk will conclude with an overview of emerging trends in the application of artificial intelligence and machine learning to healthcare and biomedical research. Key challenges, including data heterogeneity, interpretability, validation, bias, and privacy, will be highlighted, along with some caveats that need to be considered as AI increasingly becomes part of healthcare and biological discovery.
Auditorium
10:10 AM – 11:00 AM
Plenary Keynote | Dr. Ponnurangam Kumaraguru (PK)
TalkSaral AI: Democratising Scholarly Work by Transforming into Accessible, Multilingual Content
View Abstract
Have you ever considered transforming a research paper (or a patent), your own or one you are reading, into a short video, poster, slide deck, reel, podcast, business brief, or a social media post? This talk introduces Saral AI, an open-source, AI-enabled research platform designed to democratise access to research and enable large-scale knowledge diffusion. Saral AI is supported by Anusandhan National Research Foundation (ANRF). Saral AI enables researchers to convert scholarly artifacts such as research papers and patents into multiple reusable and accessible formats, including videos, slide decks, posters, reels, podcasts, and social media content. By lowering technical and communication barriers, SARAL supports researchers, educators, and institutions in amplifying the reach and impact of their work. Saral also enables researchers to consume scholarly content in multiple formats and languages, supporting inclusive and accessible knowledge engagement. A key capability of Saral AI is multilingual video generation, with outputs currently supported in Assamese, Bengali, Bodo, Dogri, English, Gujarati, Hindi, Kannada, Maithili, Malayalam, Manipuri, Marathi, Odia, Punjabi, Sanskrit, Santali, Tamil, Telugu, and Urdu, enabling research communication at population scale. The session will include a hands-on component; participants are encouraged to bring their own research paper(s) and a laptop so that they can actively use the Saral platform during the session. The talk will also cover the platform architecture, workflows, and adoption experiences. The Saral AI platform is accessible at https://saral.democratiseresearch.in/ and its open-source codebase is available at https://github.com/DemocratiseResearch/SARAL
Auditorium
11:00 AM – 11:20 AM
Tea Break
Open area in front of classroom A3-007
11:20 AM – 01:00 PM
Track Presentations (Parallel Sessions)
Track Presentations: AI 4 Cities—Enhancing Resilience, Improving Liveability
11:20 AM – 11:50 AM: Track Keynote: Dr. Nipun Batra
11:50 AM – 01:00 PM: Track Oral Presentations & Flash Talks
Track Keynote TalkAI for Cleaner Air: From Satellites to City Decisions
View Abstract
Cleaner air requires a clearer picture of pollution sources, better monitoring, and evidence that people can use. This talk presents our work at the Sustainability Lab, IIT Gandhinagar, on applying AI across these challenges. We use satellite imagery and computer vision to map brick kilns and study their evolution and regulatory compliance. We explore how active learning can guide the placement of air-quality monitoring stations, particularly where measurements are sparse. Through VayuChat, we examine how language models can make environmental data accessible through natural-language questions and executable analysis. I will also discuss ongoing work on neural emulators for faster environmental modelling and scenario exploration. These examples motivate broader questions for AI in cities: how to evaluate models under changing conditions, address uneven monitoring coverage, and turn technical advances into useful tools for researchers, citizens, and policymakers.
View Presentations & Flash Talks
Oral Presentations (15 min each: 12 min talk + 3 min Q&A)
Anand Kumar Rai
RailVaani: An Optimized Bilingual ASR Pipeline for Entity-Rich Railway Announcements
Dr. Saurabh Sharma
STARK: Enhancing Traffic Prediction Through Spatiotemporal Adaptive Refinement With Knowledge Distillation
Amal A
Decoding Customer Experience from User-Generated Content: A Multilingual Aspect-Based Sentiment Analysis Approach
Flash Talks (2 min each, 1 slide)
Partha Sarthy
Pranoy Ghosh, Ramam Agarwal, and Sohan Panda
Room: A3 - 007
Track Presentations: Recent Advances in AI
11:30 AM – 01:00 PM: Track Oral Presentations & Flash Talks
View Presentations & Flash Talks
Oral Presentation (15 min: 12 min talk + 3 min Q&A)
Anirban Majumdar
About Time: Model-free Reinforcement Learning with Timed Reward Machines
Flash Talks (2 min each, 1 slide)
Jamaluddin
Nandini Arimanda
Ravindrakumar M. Purohit
Hari Prapan
Poonam Sharma
Akshay Lakhi
Room: A3 - 009
AI 4 Cities: A3 - 007


Recent Advances in AI: A3 - 009
01:00 PM – 01:15 PM
Group Photograph
Outside Library Grand Staircase
01:15 PM – 02:10 PM
Lunch
Dining Hall (First Floor)
02:15 PM – 02:50 PM
Plaksha Conversations
A1-001(A+B)
03:00 PM – 03:50 PM
Plenary Keynote | Dr. Manish Gupta
TalkThe Transformative Power of AI and Open Challenges
View Abstract
We begin by presenting recent advances in artificial intelligence, and in particular, deep learning, foundation models, and agents. We describe the tremendous progress of these AI models on problems ranging from understanding, prediction and creativity on one hand, and open technical challenges like safety, fairness and transparency on the other hand. These challenges are further amplified as we seek to advance Inclusive AI to tackle problems for billions of human beings in the Global South. We present our work on improving multilingual capabilities and cultural understanding of foundation models like Gemini, on improving the computational efficiency of LLMs, and analyzing satellite imagery to transform agriculture and improve the lives of farmers. We then describe how powerful agentic frameworks built using these models can in turn help improve the capabilities of AI models. Through these examples, we hope to convey the excitement of the potential of AI to make a difference in the world, and also a fascinating set of open problems to tackle.
Auditorium
04:00 PM – 05:00 PM
High Tea Break / Poster Presentation
Tracks: Track 1: AI 4 Cities  |  Track 2: Recent Advances in AI  |  Track 3: Reasoning in Vision Language Models  |  Track 4: Responsible and Ethical AI
A3 First Floor Porch Area
05:00 PM – 06:15 PM
Panel Discussion
ThemeBeyond Chatbots: Bridging the Gap Between AI Use and Real-World Impact in India
Panelists: Dr. Balaraman Ravindran (IIT Madras), Dr. Shachindra Nath (Anādi Intelligence), Mr. Anoop G Prabhu (Vehant Technologies), Dr. M Balakrishnan (Plaksha University)
Moderator: Dr. Siddharth (Plaksha University)
View Panel Abstract
AI adoption in India is already growing through chatbots, coding assistants, and content-generation tools. But how do we move from individual productivity gains to solving harder problems in agriculture, healthcare, education, and small businesses, where AI must operate amid diverse languages, limited resources, unreliable data, and complex real-world conditions?

This panel will explore why AI that works well in a chat window can be difficult to translate into dependable solutions on the ground. What technical breakthroughs, better system design, or changes in deployment are needed to make AI genuinely useful beyond the screen?
Auditorium
06:30 PM – 07:30 PM
Plaksha Cultural Event
Auditorium
07:30 PM Onwards
Gala Dinner
Multipurpose Hall
Day 3: Sunday, November 01, 2026
07:00 AM – 08:30 AM
Breakfast
Dining Hall (First Floor)
08:30 AM – 09:00 AM
Refreshments
A1 - MR 1005
09:00 AM – 09:10 AM
Day 3 Opening Remarks
Auditorium
09:10 AM – 10:00 AM
Plenary Keynote | Dr. Balaraman Ravindran
TalkMeasuring What Matters: Evaluation as the Foundation of AI Governance
View Abstract
AI systems increasingly shape consequential decisions, from university choices to clinical diagnoses, yet governing them rests on an often-unexamined premise: that we can reliably measure how they behave. Benchmark accuracy reveals little about fairness, safety, explanation faithfulness, or the reliability of evaluators themselves, and poor measurement offers false assurance. This talk argues that responsible AI governance requires context-aware, multidimensional evaluation, as illustrated by recent work from the Centre for Responsible AI at IIT Madras. KANEx designs for explainability, using Kolmogorov-Arnold Networks to ground chest X-ray explanations and produce more faithful saliency maps. LLM-judge verdicts can be partially predicted from the rubric text alone, and judges often fail to update in response to counterfactual reversals. We also show that safety is fragile under everyday phrasing: exploratory framings of harmful requests cut safe-response rates by 17–21 percentage points, a gap widened by medical fine-tuning and traced mechanistically to failures in executing refusal. If time permits, I will cover two other studies - first, auditing over 25,000 LLM university recommendations, we uncover systematic geographic and gender biases, quantified through new representation scores. Second, LExT scores the trustworthiness of natural language explanations by balancing plausibility and faithfulness. Together, these works position rigorous, scrutinised measurement as the foundation for governing AI.
Auditorium
10:10 AM – 11:10 AM
Track Presentations: Reasoning in Vision Language Models
10:10 AM – 10:40 AM: Track Keynote: Dr. Vineeth N Balasubramanian- ONLINE
10:40 AM – 11:10 AM: Track Oral Presentations
View Presentations
Oral Presentations (15 min each: 12 min talk + 3 min Q&A)
Ayush Shrivastava
ThermEval: A Structured Benchmark for Evaluation of Vision-Language Models on Thermal Imagery
Kartik Bose
How Much Did the Image Matter? An Audit of Evidence Use in Medical Vision-Language Models
Auditorium
Auditorium
11:10 AM – 11:30 AM
Tea Break
Outside Auditorium
11:30 AM – 01:00 PM
Track Presentations: Responsible and Ethical AI
11:30 AM – 12:30 PM: Track Oral Presentations & Flash Talks
View Presentations & Flash Talks
Oral Presentations (15 min each: 12 min talk + 3 min Q&A)
S Aarthi
Understanding Speech, Missing Deepfake Artifacts: The Explainability Gap in Audio Language Models
Prashasti Agarwal
AI for Education Equity in Changing Higher Education: An Equity-Aware Learning Analytics Framework for India
Amit Kalita
Operationalizing fairness: A bounded threshold heuristic under hard resource limits
Sanchet Sharma
AI Personhood and the Colonial Architecture of Accountability: From Legal Fictions to Supervisory Strict Liability
Flash Talks (2 min each, 1 slide)
Rishi Ahuja
Praveen Parthasarathy
Anshumaan Goel
Dr. Gagan Jain
Rounak Mukhopadhyay
Priyanshi Tanwar
Auditorium
Auditorium
01:00 PM – 01:20 PM
Award Ceremony
Auditorium
01:20 PM – 01:40 PM
Concluding Remarks
Auditorium
01:40 PM – 01:50 PM
Group Photograph
Outside Library Grand Staircase
01:50 PM – 02:30 PM
Lunch
Dining Hall (First Floor)
02:30 PM Onwards
Chandigarh Tour & Departure
Departure for Local Tour

This website is hosted under a domain recognized by Plaksha University for departmental, research, faculty, student, and event-related digital initiatives. It is affiliated with Plaksha University but does not represent the university's official position, views, or institutional stance. Plaksha University does not review, verify, or endorse the content published on this site prior to publication and accepts no liability for its accuracy, completeness, or currency. Responsibility for the content, maintenance, and compliance of this site rests entirely with its owner, in accordance with the Plaksha University Website Hosting Policy.