Embedded Files
RC-TMLSE
Research Challenges in Tiny Machine Learning Systems at Edge
Workshop at PReMI 2025, IIT Delhi
The First International Workshop on Research Challenges in Tiny Machine Learning Systems at Edge
The First International Workshop on Research Challenges in Tiny Machine Learning Systems at Edge
(RC-TMLSE 2025)
(RC-TMLSE 2025)
AI has undergone a paradigm shift in recent years, with a recognized need for deploying DNN-based inference models on edge devices. The challenges to ensure low latency, high reliability and privacy amid potentially dynamic operating conditions and data variability/diversity are exacerbated by the device constraints and further complicated by the need for ensuring energy efficiency. Edge devices, such as smartwatches, smart home gadgets, and autonomous vehicles, are integral components of our increasingly connected world. These devices, despite their resource constraints, require advanced capabilities, including real-time data processing, learning, and decision-making. TinyML models are engineered to fit within the limited computational, memory, and energy budgets of edge devices. However, optimizing these models while maintaining accuracy and efficiency is a formidable task. In this workshop, we emphasize on the need for accurate yet lightweight and efficient machine learning models that enable diverse set of edge applications and eventually bring down the cost of intelligence. However, deploying these tiny ML models presents a host of technical challenges, necessitating innovative solutions and focused research efforts. This workshop aims to delve deep into these challenges, foster collaborative research, and drive the development of cutting-edge techniques and methodologies.
CALL FOR PAPERS
RC-TMLSE 2025: First International Workshop on Research Challenges in Tiny Machine Learning Systems at Edge
RC-TMLSE 2025: First International Workshop on Research Challenges in Tiny Machine Learning Systems at Edge
Co-located with PReMI 2025 | IIT Delhi, India | Half-Day Workshop
The RC-TMLSE 2025 workshop invites original research contributions that address the growing need for deploying efficient, accurate, and privacy-preserving machine learning models on edge devices. As edge computing becomes central to real-time, low-latency, and energy-constrained applications, this workshop aims to explore the technical challenges and innovations in TinyML systems.
Topics of Interest (include but are not limited to):
Novel TinyML model architectures
Model reduction: pruning, quantization, distillation, factorization
Foundation models at the edge (LLMs, VLMs)
Robustness to distribution shift and adversarial perturbations
Fast training and accuracy estimation with minimal data
Energy and latency proxy modeling
Dynamic resource management and task scheduling
Edge-based data compression and optimization
Continual learning and selective unlearning on-device
Federated learning in TinyML systems
Applications in embedded vision, speech, and physiological signal analytics
Target Audience
Researchers in machine learning and edge computing
Engineers working on IoT, autonomous systems, and embedded AI
Practitioners deploying TinyML in real-world scenarios
Submission Guidelines
Submission Guidelines
Submit original, unpublished research papers (up to 8 pages, excluding references)
All submissions will undergo peer review by the program committee
Format: Follow the PReMI 2025 guidelines at https://premi25.iitd.ac.in/calls/cfp.pdf
Submission link: https://openreview.net/group?id=PReMI/2025/Workshop/RC-TMLSE
Key Dates:
Technical paper submission deadline: September 30, 2025 11:59PM IST
Notification of acceptance: October 15, 2025
Camera Ready deadline: October 31, 2025
Submission Guidelines
RC-TMLSE 2025 welcomes a wide range of contributions in the areas specified in the Call for Papers. When submitting a paper to RC-TMLSE 2025, authors are required to specify one or more keywords from the list of topics outlined in the CFP. The RC-TMLSE 2025 Program Committee will endeavour to facilitate the presentation of papers from contributors worldwide.
At least one author of each paper must register for PReMI 2025 as mentioned in the PReMI guidelines.
Submissions should follow the norms, templates and guidelines of PReMI 2025.
Reviews will be double-blind. Authors should maintain this during submission of the paper.
Workshop papers will be of 8 pages (maximum) and must use the template given by the conference organizers.
Papers will be submitted via OpenReview. Each author must create a profile at OpenReview for submissions. Please note that as per OpenReview policy, new profiles created without an institutional email will go through a moderation process that can take up to two weeks and new profiles created with an institutional email will be activated automatically.
Paper submission link: https://openreview.net/group?id=PReMI/2025/Workshop/RC-TMLSE
Organizing Committee
Organizing Committee
Utpal garain
Utpal garain
Utpal Garain is a Professor at the Indian Statistical Institute, Kolkata, specializing in deep learning for language, image, and video analysis. He received his B.E. and M.E. in Computer Science and Engineering from Jadavpur University and earned a Ph.D. from the Indian Statistical Institute in 2005. He began his career in the software industry before transitioning to academia. His research spans OCR, handwriting analysis, NLP, and computational forensics, with strong ties to Indian language technologies (TDIL). Prof. Garain has held prestigious fellowships including a postdoctoral stint in France (CNRS), the Indo-US Research Fellowship (IUSSTF, 2011), and the JSPS Invitational Fellowship in Japan (2016). He serves as Associate Editor for IJDAR and chairs IAPR TC-6 on Computational Forensics. He regularly contributes to top conferences like ICPR and ICDAR and reviews for journals including IEEE PAMI and PR. His contributions earned him the INAE Young Engineer Award in 2006.
Shiwei Liu
Shiwei Liu
Shiwei Liu is an incoming Principal Investigator at the ELLIS Institute Tübingen and a Group Leader at the Max Planck Institute for Intelligent Systems. He is currently a Royal Society Newton International Fellow at the University of Oxford and a Junior Research Fellow at Somerville College. Previously, he was a postdoctoral fellow in the VITA group at UT Austin, working with Atlas Wang, funded by IFML. Shiwei earned his Ph.D. in Computer Science from Eindhoven University of Technology (TU/e), the Netherlands, under the supervision of Mykola Pechenizkiy and Decebal Constantin Mocanu.
His research focuses on understanding and harnessing low-dimensional structures in machine learning. This perspective informs efficient training and inference of foundation models, advances reasoning in large language models, and drives the development of hardware-friendly ML algorithms. His work sits at the intersection of theoretical insight and scalable, resource-aware AI design.
Arpan Pal
Arpan Pal
Dr. Arpan Pal is a Distinguished Chief Scientist and Research Area Head at TCS Research, leading Embedded Devices and Intelligent Systems. With over 30 years of experience, his expertise spans intelligent sensing, signal processing, AI at the edge, and affective computing, with impactful work in Connected Health, Smart Manufacturing, Retail, and Remote Sensing. He has filed over 180 patents (95+ granted globally) and published 160+ papers and book chapters, along with three books on IoT, digital twins, and AI in cardiac screening.
Dr. Pal serves on editorial boards of ACM TECS and Springer’s Journal of Computer Science, and program committees of IEEE Sensors, ICASSP, and EUSIPCO. He advises government bodies like CSIR and MeitY, and academic institutions including IITs and IIITs. A two-time Tata InnoVista innovation awardee, he previously worked at DRDO and Rebeca Technologies. He holds a Ph.D. from Aalborg University and B.Tech/M.Tech from IIT Kharagpur.
Swarnava Dey
Swarnava Dey
Swarnava Dey is a Senior Scientist at TCS Research with over 24 years of industrial research experience in embedded systems, spanning computer vision, sensor analytics, speech processing, and edge AI. He holds an M.Tech from IIT Kharagpur and is currently pursuing a Ph.D. there. At TCS Research, he leads AutoML efforts within the Embedded Devices and Intelligent Systems team, focusing on Neural Architecture Search and model reduction for tinyML applications in health, manufacturing, retail, and remote sensing. Swarnava has filed 80+ patents (35+ granted globally) and published over 40 papers in top-tier venues. He was a winner of IEEE BigData Hackathon (2018) and IJCNN AI Competition (2023), and is Google-certified in ML and Cloud. A regular reviewer for CVPR, ICCV, TPAMI, IJCV, and major AI journals & conferences, he is among the few researchers exploring adversarial and OOD robustness in compact vision models.
Gitesh Kulkarni
Gitesh Kulkarni
Gitesh Kulkarni is working as Senior Scientist in TCS research. He is an accomplished designer of embedded systems design and a Maker at heart. He is Master of Science in Electrical engineering from Colorado State University, Fort Collins, Colorado, USA. His research interests are at the cusp of edge computing, computer architectures, and sensing. In one of pioneering works, he was the lead designer of the first industrial safety watch for TATA sons. Prior to joining TCS, Gitesh created embedded systems and systems products for leading technology companies. When not playing with Linux and soldering gun, he is busy reading novels, comics, and musing about philosophy. He is a licensed Ham radio operator as well.
Arijit Ukil
Arijit Ukil
Dr. Arijit Ukil (Senior Member, IEEE) has over 22 years of industrial research experience. He is currently serving as a Principal Scientist at TCS Research, Tata Consultancy Services. He previously worked at the Defence Research and Development Organization (DRDO) as Scientist 'C'. He holds a master’s degree in engineering from Jadavpur University, Kolkata. He holds PhD (Cum Laude) from University of Murcia, Spain.
Dr. Ukil’s research spans deep learning, Gen Ai, causal AI, and model optimization. He has authored over 60 publications, four book chapters, and filed 50+ patents, with 40+ patents granted globally across regions including the US, EU, China, and Japan. He actively promotes edge AI, having spoken at the tinyML forum on ECG classification for wearable and implantable devices. He was a keynote speaker at IEEE AINA-2016, panel moderator at CIKM 2020, and serves as Steering Committee Member of HealthyIoT. He served as the general chair in KDAH workshop of ACM CIKM and ICASSP 2024 workshop of Deep Neural Network Model Compression. He delivered tutorials on edge analytics and deep model compression in top conferences like IEEE ICASSP (2024, 2025), ACM SAC (2025).
Raunaque Mujeeb Quaiser
Raunaque Mujeeb Quaiser
Raunaque is currently working with STMicroelectronics and brings a total of 21 years of experience in the semiconductor industry. He holds a B. Tech in Computer Science from AMU, an MBA from IIM Lucknow, and is currently pursuing a PhD in Innovation Strategy at IIM Rohtak. His key responsibilities include electronic system development across various technologies, with a focus on augmented reality (AR) and smart glasses. His areas of interest encompass connectivity technologies, mobile applications, as well as system development and integration. Additionally, he manages the ST for Startups program for the India region and actively collaborates with universities on initiatives such as curriculum development and embedded labs.
Arijit Mukherjee
Dr. Arijit Mukherjee, a Principal Scientist at TCS Research, specializes in embedding intelligence at the edge, focusing on low-power, low-latency computing and brain-inspired neuromorphic processing. After earning Bachelors and Masters degrees in Computer Science, Arijit worked in the software industry before joining Newcastle University, UK, where he worked as a lead researcher in several UK eScience projects contributing to the development of W3C standards for Web Services/SOA and the concept of Cloud. With a PhD in Grid Computing, he returned to Kolkata, worked in a telecommunications revenue assurance product company for three years before joining TCS Research in 2011. He currently leads research in Edge and Neuromorphic Systems. A Distinguished Scientist awardee at TCS Research, Arijit has 75+ publications and 50+ patents across multiple geographies.
Venue
Venue
IIT Delhi
IIT Delhi
Hauz Khas, New Delhi-110016, INDIA
Indian Institute of Technology Delhi is one of the 23 IITs created to be Centres of Excellence for training, research and development in science, engineering and technology in India. Established as College of Engineering in 1961, the Institute was later declared as an Institution of National Importance under the “Institutes of Technology (Amendment) Act, 1963” and was renamed as “Indian Institute of Technology Delhi”. It was then accorded the status of a Deemed University with powers to decide its own academic policy, to conduct its own examinations, and to award its own degrees. Since its inception, over 60000 have graduated from IIT Delhi in various disciplines including Engineering, Physical Sciences, Management and Humanities & Social Sciences. Of these, nearly 5070 received Ph.D. degrees. The number of students who graduated with B.Tech. degree is over 15738. The rest obtained Master’s Degree in Engineering, Sciences and Business Administration. These alumni today work as scientists, technologists, business managers and entrepreneurs. There are several alumni who have moved away from their original disciplines and have taken to administrative services, active politics or are with NGOs. In doing so, they have contributed significantly to building of this nation, and to industrialization around the world.
Indian Institute of Technology Delhi is one of the 23 IITs created to be Centres of Excellence for training, research and development in science, engineering and technology in India. Established as College of Engineering in 1961, the Institute was later declared as an Institution of National Importance under the “Institutes of Technology (Amendment) Act, 1963” and was renamed as “Indian Institute of Technology Delhi”. It was then accorded the status of a Deemed University with powers to decide its own academic policy, to conduct its own examinations, and to award its own degrees. Since its inception, over 60000 have graduated from IIT Delhi in various disciplines including Engineering, Physical Sciences, Management and Humanities & Social Sciences. Of these, nearly 5070 received Ph.D. degrees. The number of students who graduated with B.Tech. degree is over 15738. The rest obtained Master’s Degree in Engineering, Sciences and Business Administration. These alumni today work as scientists, technologists, business managers and entrepreneurs. There are several alumni who have moved away from their original disciplines and have taken to administrative services, active politics or are with NGOs. In doing so, they have contributed significantly to building of this nation, and to industrialization around the world.
Page updated
Google Sites
Report abuse