On-Device Training

Follow this topic Ask about this topic

Momentum

4 papers in the last four weeks, against 2 the four weeks before. 0.0% of all new papers.

Jul 13Week of Sep 28

Latest papers 21

All topics

MemFLoRA: Memory-Floor LoRA for CNN Adaptation at the Edge
Oct 6, 2026Mehmet Emre Akbulut, Johannes Geier, Ulf SchlichtmannOn-Device Training Memory-Efficient Fine-Tuning
Technical University of Munich Munich, Germany
Lock-in EP: An In-Situ Training Algorithm for Oscillatory Hardware
Oct 5, 2026Sowjanya Tammali, Wilkie Olin-AmmentorpEquilibrium Propagation On-Device Training
Computer Science, Missouri University of Science and Technology, 500 West 15th Street, Rolla, 65409, MO, USA. · Mathematics & Computer Science Division, Argonne National Laboratory, 9700 S. Cass Avenue, Lemont, 60439, IL, USA.
Fine-Tuning a 3B-Parameter LLM on a Smartphone: Characterizing Sustained Training
Oct 5, 2026Andrew Geyko, Marius Mosbach, André BrinkmannOn-Device Training LLM Fine-Tuning
Saarland University Saarbrücken, Germany
Few-Shot Prototype Head Adaptation for On-Device ECG Personalization on PSoC~6
Oct 5, 2026Guilherme Silva, Pedro Silva, Gladston Moreira +1Prototype-Based Learning On-Device Training
Computing Department, Universidade Federal de São João del-Rei – São João del-Rei, Brazil · Computing Department, Federal University of Ouro Preto – Ouro Preto, Brazil
Towards a Cloud Fog Edge System for Smart Building
Oct 1, 2026Christophe Cérin, Mamadou Sow, Frédéric AndrèsOn-Device Training Edge ML
University Sorbonne Paris Nord LIPN, UMR CNRS 7030 INRIA DATAMOVE Paris/Grenoble, France · University Sorbonne Paris Nord LIPN, UMR CNRS 7030 Villetaneuse, France · National institute of Informatics 2 Chome-1-2 Hitotsubashi, Chiyoda City, Tokyo 101-8430, Japan
HO-FL: Hybrid-Order Federated Learning for Heterogeneous Edge Devices
Sep 30, 2026Qiyuan Chen, Xian Wu, Yanan Ma +1On-Device Training Federated Learning
University of Hong Kong · City University of Hong Kong
TinyUDE: Solver-Free Universal Differential Equations on Microcontrollers via Lie-Taylor Jet Matching
Sep 22, 2026Pranavanath Balamurali, Hrishi KamireddyOn-Device Training System Identification
Department of Electrical and Computer Engineering The University of Texas at Austin · Department of Computer Science The University of Texas at Austin
TASTE: Throughput-Aware Batch Size Tuning for On-Device Edge Learning
Sep 7, 2026Avik Bhatnagar, Federico Nicolas Peccia, Oliver BringmannOn-Device Training Edge ML
FZI Research Center for Information Technology, University of Tübingen Germany
Curvature-Aware Zeroth-Order Optimization for Memory-Efficient Test-Time Adaptation
Aug 12, 2026Junming Zhang, Shuyu Yin, Peilin Liu +2On-Device Training Test-Time Adaptation
Shanghai Jiao Tong University, China
Unifying Physical Backpropagation
Aug 12, 2026Cyrill Bösch, Yigithan Gediz, Clara C. Wanjura +1On-Device Training Backpropagation
Department of Computer Science, Princeton University, Princeton, NJ 08540, USA · Department of Electrical and Computer Engineering, Princeton University, Princeton, NJ 08540, USA
ZeroLock: Concurrent Memory-Efficient LLM Training via Modular Update Decoupling
Aug 8, 2026Wentao Dai, Xuanran Li, Yuxiang Zhang +2On-Device Training Memory-Efficient Fine-Tuning
Department of Computer Science and Engineering, Southern University of Science and Technology, China · Department of Mathematics, Southern University of Science and Technology, China · School of Computing, Montclair State University, New Jersey, USA
MAUPITI: On-Device Prototype-Based Learning on a Smart Infrared Sensor
Aug 7, 2026Beatrice Alessandra Motetti, Tanguy Dugas du Villard, Matteo Risso +7Prototype-Based Classification On-Device Training
Politecnico di Torino, Turin, Italy · STMicroelectronics, Cornaredo, Italy
Design-Time Optimization of Deep Neural Networks for Intermittent Learning on Microcontrollers
Aug 4, 2026Jakob Schubert, Maximilian Kasper, Maximilian Linke +5On-Device Training Energy-Efficient ML
Fraunhofer Institute for Integrated Circuits IIS, Nürnberg, Germany · Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU), Erlangen, Germany · Ruhr University Bochum (RUB), Bochum, Germany +2
Versatile On-device Adaptation at the Edge by Unifying Few-shot, Zero-shot, Continual, and In-context Learning
Jul 31, 2026Douwe den Blanken, Martin Lefebvre, Charlotte FrenkelZero-Shot Learning On-Device Training
Microelectronics Department (EEMCS Faculty), Delft University of Technology, 2628 CD Delft, Netherlands
Hardware-Software Co-Design for Float16 On-Device Training on RISC-V Single-Core
Jul 23, 2026Benjamin Hubinet, Pierre-Alain Moellic, Olivier Savry +2On-Device Training Hardware-Software Co-Design
CEA-Leti, Mines Saint-Etienne, Equipe Commune SAS, F-13541 Gardanne, France · Univ. Grenoble Alpes, CEA-Leti, F-38000 Grenoble, France · Mines Saint-Etienne, CEA-Leti, Centre CMP, Equipe commune SAS, F-13541 Gardanne, France
QScheduler: Adaptive Gradient Sampling for Zeroth-Order On-Device Training on INT8 NPUs
Jul 21, 2026Victor Felipe Domingues Do Amaral, Pierre Demaj, Erwan Libessart +3On-Device Training Zeroth-Order Optimization
GeePs · STMicroelectronics, France · GeePs, CNRS, CentraleSup´elec, Universit´e Paris-Saclay, Sorbonne Universit´e, France
Empowering On-Device Model Adaptation with an Edge AI Inference Accelerator
Jul 20, 2026Mateusz Piechocki, Alessandro Capotondi, Marek KraftOn-Device Training Edge ML
Poznan University of Technology, Institute of Robotics and Machine Intelligence, Poznan, Poland · University of Modena and Reggio Emilia, The High-Performance Real-Time Laboratory, Modena, Italy
On-Device Adaptive Battery Power Prediction for Electric Vehicles
Jul 10, 2026Avik Bhatnagar, Anton Paule, Tobias Schuermann +2On-Device Training Continual Test-Time Adaptation
FZI Research Center for Information Technology, University of Tübingen Germany
What changes after deployment? A survey on On-device Learning in TinyML
May 29, 2026Massimo Pavan, Luca Pezzarossa, Fabrizio Pittorino +2On-Device Training Distribution Shift
Technical University of Denmark (DTU), Kongens Lyngby, Denmark · Politecnico di Milano, Milan, Italy
Spiker-LL: An Energy-Efficient FPGA Accelerator Enabling Adaptive Local Learning in Spiking Neural Networks
May 18, 2026Alessio Caviglia, Filippo Marostica, Alessandro Savino +1On-Device Training Energy-Efficient ML
Control and Computer Engineering Department, Politecnico di Torino, Turin, Italy
On-Device Vision Training, Deployment, and Inference on a Thumb-Sized Microcontroller
Apr 24, 2026Jeremy EllisOn-Device Training Tiny ML
High School Robotics Educator, British Columbia, Canada

Oct 6, 2026, cs.LGJ/K move · Enter open · S save

MemFLoRA: Memory-Floor LoRA for CNN Adaptation at the Edge

Mehmet Emre Akbulut, Johannes Geier, Ulf Schlichtmann

Technical University of Munich Munich, Germany

On-Device Training Memory-Efficient Fine-Tuning Domain Adaptation Low-Rank Adaptation Convolutional Neural Networks

Open paper AskPDF arXiv

On-device learning is necessary when the model encounters user-,sensor-, or environment-specific shifts after deployment. Although parameter-efficient fine-tuning (PEFT) methods, particularly Low-Rank Adaptation (LoRA) variants, enable efficient adaptation at the edge, the limiting resource for Convolutional Neural Network (CNN) adaptation is often not the number of trainable parameters but the activation state that must be retained until the backward pass. This paper introduces Memory-Floor LoRA (MemFLoRA), a low-rank CNN adapter built around a memory-first design principle rather than a direct application of transformer-oriented LoRA. Instead of merely reducing trainable weights, we define an activation-memory-floor criterion: trainable backward computations must not depend on full-width layer inputs. The resulting adapter freezes the down-projection, trains a scale-matched up-projection, and combines eval-mode backbone normalization with activation-minimal backward rules, reducing saved state to the low-rank branch. Evaluated on three Human Activity Recognition (HAR) datasets and two CNN backbones under subject, body-location, and sensor-placement shifts, MemFLoRA reduces saved-activation memory by 98.5-98.7% and peak training-state memory by 94.9-97.3% relative to full fine-tuning, while matching or exceeding CNN PEFT baselines.

Momentum

Related topics

Latest papers 21