Skip to yearly menu bar Skip to main content

Workshop Sat, Dec 12, 2026 • 8:00 AM – 5:00 PM AEDT MR C3.6

On-Device Intelligence: Foundation Models under Real-World Constraints

Niao He ⋅ Bingcong Li ⋅ Shiwei Liu ⋅ Hao Ma ⋅ Michael Muehlebach ⋅ Daniela Rus ⋅ Marko Zaric ⋅ Melanie Zeilinger

[ OpenReview]

Abstract

Intelligence should not rely solely on the cloud. As foundation models enter real-world interactive and embodied settings, their cloud-centric capabilities must be realized on devices under constraints on latency, memory, reliability and other factors. From LLMs to embodied foundation models, the key question is no longer only how to improve capability, but how to make it usable for efficient on-device inference and reasoning, stable interaction, continual adaptation, and safe execution. On-device intelligence is therefore not merely model compression or deployment, but a tightly coupled research problem spanning theory, algorithms, systems, and evaluation, where capability must be considered jointly with efficiency, adaptation, execution, and reliability under real-world constraints. This workshop will bring together researchers working on LLMs, embodied intelligence, and edge systems to clarify this shared problem space and identify key questions through invited talks, panels, contributed posters, and open discussions, helping connect separated communities and shape a shared research agenda for on-device intelligence.

Chat is not available.

Schedule

Time zone: Conference local time zone

StreamSE: Neural Audio Codec-based Streaming Speech Enhancement for On-device Inference

Hyunjin Choi ⋅ Zinu Kim ⋅ Junwon Moon ⋅ Jeongyun Chae ⋅ Seungbeom Kim ⋅ Kyuhong Shim

LFM2.5 Encoders: On-Device Retrieval, Reranking, and Routing for Agentic Workflows

Edoardo Mosca ⋅ Fernando Neto Fernandes ⋅ Jens Luecke ⋅ Leonie Monigatti ⋅ Sanghyuk Lee ⋅ Rouzbeh Amin Tafreshi ⋅ Maxime Labonne

A Coherent Encoder for Q2\_0, and the Limits of Free Permutations

Grant C Forbes ⋅ Sriman V Donthireddi ⋅ Jonathon Romero ⋅ Zixiang Tang ⋅ Adrian Chan ⋅ Jianbang Zhang

AffordVLA: Implicit Affordance Representation Alignment for Vision-Language-Action Models

Weijie Kong ⋅ Yi Ma

FedSym: A Symbolic Framework for Communication-Efficient Federated Learning

Jing X Tan ⋅ Abhirup Ghosh ⋅ Soumya Banerjee

Eliminating Nondeterminism in LLM Inference via Integer-Only Arithmetic

Nathan Barry

DexHoldem: Playing Texas Hold'em with Dexterous Embodied System

Feng Chen ⋅ Tianzhe Chu ⋅ Li Sun ⋅ Pei Zhou ⋅ Zhuxiu Xu ⋅ Shenghua Gao ⋅ Simon Zhai ⋅ Yanchao Yang ⋅ Yi Ma

Models that Miss Together: Dependence-Aware Conformal Routing for Language Models

Siva Rajesh Kasa ⋅ Karan Gupta

MemEdge: Dual-Temperature Edge-Cloud Orchestration for On-Device Persistent Memory

Jifu Li ⋅ Xueshu Chen ⋅ Shiming Wang ⋅ Zhen Bi ⋅ Jungang Lou

Intelligence per Watt: Measuring Intelligence Efficiency of Local AI

Jon Saad-Falcon ⋅ Avanika Narayan ⋅ Hakki Akengin ⋅ J. W Griffin ⋅ Herumb Shandilya ⋅ Adrian G Lafuente ⋅ Medhya Goel ⋅ Rebecca Joseph ⋅ Shlok Natarajan ⋅ Etash Guha ⋅ Shang Zhu ⋅ Ben Athiwaratkun ⋅ Azalia Mirhoseini ⋅ Christopher Ré

Towards Lightweight Visual Routing for Mobile Agents via Fine-Grained Screenshot Synthesis

Cheng-Che Lee ⋅ Chia-Pin Cheng ⋅ Hsu-Sheng Chang ⋅ CHIALIN CHOU

Revisiting the Shape Convention of Transformer Language Models

FengTing Liao ⋅ Guan-Ting Yi ⋅ Tzu-Quan Lin ⋅ Meng-Hsi Chen ⋅ Da-shan Shiu

Below One Bit: Training Route and Budget Shape Robustness Under On-Device Storage Constraints

Yash Maheshwari

History-Augmented Temporal Attention for Waypoint Handoff in NMPC Policy Distillation

Nicholas Huang ⋅ Parvel Gu

LoopSpec: Pipelined Self-Speculative Decoding for Looped Transformers

SangLyul Cho ⋅ Langqing Cui ⋅ Insu Han

StallBound: The Memory-Bandwidth Limit of Chunked Prefill

Ahnaf Adib ⋅ Sabbir Ahmed

Agentic RF Intelligence: Multi-Timescale 6G Sensing and Reasoning with On-Device Foundation Models

Jaron Fontaine ⋅ Jelle De Moerloose ⋅ Xander Vanparys ⋅ Anton Lambrecht ⋅ Eli De Poorter ⋅ Adnan Shahid

Which Tokens Can a Device Afford to Score? Vocabulary Pruning by Log-Mass Maximization

Sergey IVANOV ⋅ Egor Krasheninnikov ⋅ Nikita Kozodoi ⋅ Natalia Tarasova ⋅ Aoyu Zhang

A Neighbor-based Expansion Approach to Graph Federated Recommendation

Aymen R Khouas ⋅ Mohamed Reda Bouadjenek ⋅ Hakim Hacid ⋅ Sunil Aryal

GPR: Grounding-Preserving Recovery for Quantized Vision-Language Models in Edge Deployments

Eric Liang ⋅ Warren Yang ⋅ Amogh Gupta

Certified Continuation and the Contraction Spectrum: Output-Certified Head–Block Routing for Long-Context Decoding

Shriraghav Ashok ⋅ Moses Lua

Parameter Scope Matters: MNE-L2 for Reliable Low-Latency ANN-to-SNN Conversion

Sicheng Liu ⋅ Ke Wang ⋅ Yizong Xing ⋅ Jiao Lin ⋅ Estrid He

Geometry-Preserving Inference for Multi-Agent Motion Models

Ryan Hardesty Lewis

Spatial Temporal Reasoning Models for Accurate and Efficient Vision-Based Localization

Jeffrey Krichmar ⋅ Hin Wai Lui

Efficient and Robust Drafting with Multi-Token Prediction

Filippo Galgani ⋅ Ziwei Ji ⋅ Omri Homburger ⋅ Ganesh Jawahar ⋅ Michelle Tadmor Ramanovich ⋅ Ivan Korotkov ⋅ Luciano Martins ⋅ Utku Evci ⋅ Edouard YVINEC ⋅ Tal Schuster

AlterNet: Adaptive-Depth Mixture-of-Experts with Learned Halting for Elastic Inference

Alexander Kastius

Robust Parameter-Efficient LLM Adaptation on Analog Hardware

Jindan Li ⋅ Zhaoxian Wu ⋅ Tianyi Chen

FirmVLA: Sequence-SLO Serving for Heterogeneous VLA Fleets

Ahnaf Adib ⋅ Sabbir Ahmed ⋅ Latifur R Khan

The Prefill Ceiling: Quantizing the KV Cache Shortens Usable Context on Apple Silicon

Alex Huang ⋅ Wei Niu

Self-Growing On-Device AI Model Expanding Parametric Memory Without Rewriting Factual Knowledge

Aayan Arish ⋅ Yilong Li ⋅ Suman Banerjee

Reconfigurable On-Chip AI for Particle Physics Detectors with Logic Neural Networks

Mila Bileska ⋅ Lino Gerlach ⋅ G.J. Peter Elmer ⋅ Isabel R Ojalvo ⋅ Jennet Dickinson ⋅ Lindsey A Gray ⋅ Giuseppe Di Guglielmo

The Recurrent Transformer: Greater Effective Depth and Reduced KV-Cache Footprint

Costin-Andrei Oncescu ⋅ Depen Morwani ⋅ Samy Jelassi ⋅ Alexandru Meterez ⋅ Mujin Kwun ⋅ Sham Kakade

OpenJarvis: Personal AI, On Personal Devices

Jon Saad-Falcon ⋅ Avanika Narayan ⋅ Robby Manihani ⋅ Tanvir Bhathal ⋅ Herumb Shandilya ⋅ Hakki Akengin ⋅ Gabriel Bo ⋅ Andrew Park ⋅ Matthew Hart ⋅ Caia Costello ⋅ Chuan Li ⋅ Christopher Ré ⋅ Azalia Mirhoseini

Same Policy, Different Artifacts: Paired Task Audits for On-Device Neural Control

Chanseok Kang

UniCache: Adaptive Task- and Type-Aware KV Cache Compression for Unified Multimodal Models

Wanqi Yang ⋅ Yuexiao Ma ⋅ Mei Xie ⋅ Xiawu Zheng

On-Device FoRA: Native Device Fine-Tuning with Selective Riemannian Adaptation

Juneyoung Park ⋅ Seongbae Lee ⋅ Han-Sang Lee ⋅ Kyuho Lee ⋅ Minjae Kim ⋅ Seungheon Hyeon ⋅ KIDUK KWON ⋅ Seongwan Kim ⋅ Jaeho Lee

Video Token Compression via Formation and Selection Distillation in Large Vision Language Models

Soowon Son ⋅ Heeseong Shin ⋅ Jihwan Eom ⋅ Jaeyeol Jeon ⋅ Sunghun Kang ⋅ Woo Young Kang ⋅ Seungryong Kim ⋅ Byungseok Roh

Where Edge VLA Latency Actually Goes: Measured Scaling Laws and Staleness Admissibility for a 450M Vision–Language–Action Model

Prarabdh Misra ⋅ Mohd Ariful Haque

Opt.Gear: Designing Hybrid Language Models for Device-Native Inference

Juneyoung Park ⋅ YOUNGWOOK KWON ⋅ Seongwan Kim ⋅ Jaeho Lee

Compression has an accent: aggregate PTQ benchmarks hide who pays for on-device speech recognition

Bruce Peters ⋅ Diego Mardian

Which Transformer Blocks Should You Keep? An Exhaustive Map of Layer-Subset Selection in a Pretrained ViT

Steffen Jung ⋅ Margret Keuper

Evaluating the Applicability of LLM Inference Optimizations on Apple Silicon

Kyrylo Yemets ⋅ Andrii Yaroshevych ⋅ Victor Muryn ⋅ Ostap Khomenko ⋅ Pavlo Kryven ⋅ Maksym Kmet ⋅ Maksym Shamrai ⋅ Mariya Hirna

Memory-based Self-Explainable Early Exit Networks

Evan Kaiden ⋅ Leilani Gilpin ⋅ Biagio La Rosa

Same Text, Different Prediction: Serving-Context Nondeterminism in Text Classifiers

SANTHOSH KASA ⋅ Siva Rajesh Kasa ⋅ Sumit Negi

Llama-Mobile: Efficient 2.7-Bit Quantization of VLMs

Luka Ribar ⋅ Jeevan Bhoot ⋅ Douglas Orr

SLORR: Simple and Efficient In-Training Low-Rank Regularization

David González-Martínez ⋅ Shiwei Liu

When Does Fine-Tuning Overtake Prompting? An Accuracy and Cost Analysis of Small Language Models on Narrow Tasks

Urs Spiegelhalter ⋅ Jörg Franke ⋅ Frank Hutter

Efficient Local Vocabulary Selection for Context-Aware Augmentative Communication

Rishi Karumanchi ⋅ Tanay Vinaykya

PHASEGATE: Phase-Aware CPU Retrieval Scheduling for On-Device LLMs on Unified Memory

Seoyoon Yum ⋅ Sehoon Kim

When Load-Balancing Goes Too Far: Expert Pruning in Over-Dispersed Mixture-of-Experts Models

Berkcan Kapusuzoglu ⋅ Connor Pryor ⋅ Sangwoo Cho ⋅ Supriyo Chakraborty ⋅ Shi-Xiong Zhang ⋅ Sambit Sahu ⋅ Milind Naphade

The Gap Is the Loop: What Background Work Really Costs On-Device LLM Decode on Apple Silicon

Alex Huang

GLIMPSE: Real-Time Text Recognition and Contextual Understanding for VQA in Wearables

Akhil Ramachandran ⋅ Ankit Arun ⋅ Ashish Shenoy ⋅ ABHAY HARPALE ⋅ Srihari Jayakumar ⋅ Debojeet Chatterjee ⋅ Mohsen Moslehpour ⋅ Pierce Chuang ⋅ Yichao Lu ⋅ Vikas Bhardwaj ⋅ Peyman Heidari

The Optimal Brain Size for a Body

Baihan Lin

Where Privacy Noise Belongs in On-Device AI: Rare-Signal Measurement in Conversational Advertising

Dipankar Sarkar

Group Think: Collaborative Parallel Reasoning Model

FengTing Liao ⋅ Davide Buffelli ⋅ Chan-Jan Hsu ⋅ Jamie McGowan ⋅ Sattar Vakili ⋅ Yi-Chang Chen ⋅ YI W WU ⋅ Wen-Tse Hsu ⋅ Da-shan Shiu

Tap2Listen: Instant Target Switching for Streaming Audio-Visual Speaker Extraction

Hansol Park ⋅ Minkyung Song ⋅ Zinu Kim ⋅ Yu J Lee ⋅ Hoseong Ahn ⋅ Hyunjin Choi ⋅ Kyuhong Shim

Linear-Time Non-Local Codec Placement for Machine Learning Model Compression

Matteo Spallanzani

Whole-Model Quantization Below 4 Bits Per Weight for On-Device Multimodal Models

Karen Mosoyan ⋅ Henry Ndubuaku ⋅ Jakub Mroz ⋅ Noah Cylich ⋅ Roman Shemet ⋅ Parkirat Sandhu ⋅ Justin H Lee ⋅ Satyajit Kumar

ixi-sLM: On-Device sLM for AI Call Agent Service

Han-Sang Lee ⋅ Kyuho Lee ⋅ Jin K Park ⋅ Minjae Kim ⋅ Seungheon Hyeon ⋅ KIDUK KWON ⋅ Juneyoung Park ⋅ Seongbae Lee ⋅ Yuri Hong ⋅ Jinwoo Lee ⋅ YOUNGWOOK KWON ⋅ Seongwan Kim ⋅ Jaeho Lee

Agents unlock new capabilities through Switching LoRA Adapters as a Tool (SLAaaT)

Kenneth Ge

Progressive UAV Video Token Communication for Bandwidth-Limited Wireless Transmission

Qi Qiu ⋅ Zhongqiang Zhang ⋅ GUANGMING Shi ⋅ Shaohua Wu ⋅ Ye Wang ⋅ Shuai MA ⋅ Bin Li

One Loss Is Not Enough: Cross-Objective Structured Pruning for Reasoning Vision–Language–Action Models

Seongchan Kim ⋅ Soowon Son ⋅ Sangoh Kim ⋅ Hyunah Ko ⋅ Eunbeen Hong ⋅ Hyunkoo Lee ⋅ Sanghyun Lee ⋅ WonJun Moon ⋅ Kihong Kim ⋅ Dongyeon Son ⋅ Woongchan Nam ⋅ Seungryong Kim

EfficientXpert: Efficient Domain Adaptation for Large Language Models via Propagation-Aware Pruning

Songlin Zhao ⋅ Michael Pitts ⋅ Zhuwei Qin

Quantization-Aware Distillation for 4-Bit On-Device Language Models

Aditya Tadimeti ⋅ Anthony Maltsev

Improving Efficient Recurrent Depth Via Next-Latent Alignment

Daniel Tcheurekdjian ⋅ Yuekun Yao ⋅ Huan Sun

Right Tool, Wrong Arguments: Permission Drift in Quantized Tool-Calling LLMs

Sivadinesh Ponrajan ⋅ Monik Raj Behera ⋅ Thorsteinn S Rögnvaldsson ⋅ Prayag Tiwari

Multi-Resolution Flow Matching: Training-Free Diffusion Acceleration via Staged Sampling

Xingyu Zheng ⋅ Xianglong Liu ⋅ Yifu Ding ⋅ Weilun Feng ⋅ Junqing Lin ⋅ Jinyang Guo ⋅ Haotong Qin

Multi-Mixer Models: Flexible Sequence Modeling with Shared Representations

Kevin Li ⋅ Asher Trockman ⋅ Ananda Theertha Suresh ⋅ Ziteng Sun

ElasticInfer: Safe Contextual Model Routing under Dynamic Edge-Resource Constraints

Snehasis Banerjee ⋅ Madhava Krishna

A Study of Accelerating Flow-Based VLA Inference via a Plug-and-Play Episodic Cache

Ryuji Oi ⋅ Hikari Otsuka ⋅ Kosuke Matsushima ⋅ Yuki Ichikawa ⋅ Masato Motomura ⋅ Tatsuya Kaneko ⋅ Daichi Fujiki

PROTEUS: Agentic Microarchitecture Search for FPGA Language-Model Accelerators

Andy Dimnaku ⋅ Tun-Yu Chang ⋅ Qizheng Zhang ⋅ Genghan Zhang ⋅ Rubens Lacouture ⋅ Konstantin Hossfeld ⋅ Kunle Olukotun

Confidently Wrong, Silently So: Auditing Undetectable Failures of a Deployed On-Device Language Model

Shashwat Pandey ⋅ Satwik Pandey ⋅ Suresh Raghu

Turnip: Tokenizers Are Secretly Context Compressors for Transformers

Yongchang Hao ⋅ Sepidehsadat Hosseini ⋅ Mohamed Osama Ahmed ⋅ Fred Tung ⋅ Lili Mou

Knowledge Distillation from a Raw-Signal Foundation Model to a Feature-Based On-Device Sleep Stager

Lucas F Buzuti ⋅ Daniel I Futata ⋅ Jhonatas Conceição ⋅ Felipe Chaud Pinheiro ⋅ Jean P de Matos ⋅ Byeong-ho Lee ⋅ Seungman Yang ⋅ William Camilo Ariza-Zambrano

Temporal Mixture-of-Experts: Serving Memory that Scales with Active Parameters

Noah Cylich ⋅ Mohsen Fayyaz ⋅ Henry Ndubuaku ⋅ Nanyun Peng

Extreme Low-Bit Inference in Reasoning Models: Failure Modes and Targeted Recovery

Ekaterina Alimaskina ⋅ Darya Rudas ⋅ Denis Shveykin ⋅ Gleb Molodtsov ⋅ Pavel Vasiliev ⋅ Aleksandr Beznosikov

Input-Adaptive Gating of a Dehazing Front-End for On-Device Perception in Smoke-Obscured Environments

Ishaan Garg ⋅ Seongjun Kang ⋅ Vishnu Bharadwaj ⋅ Anish Vuppala

On-Device Omni-Modal Search on Apple Silicon

Han Xiao

DualVLN on Jetson Thor: Measuring Inference Cost, Observation Freshness, and Navigation Quality

Yoojin Lim ⋅ Choonghan Lee ⋅ Yongseung Yu ⋅ Jinse Kwon

DCM-SAM: Defect-Conditioned Mixture of LoRA Experts for NPU-Deployed AM Defect Segmentation

Md Mushfiqur Rahaman ⋅ Mahedi Hasan ⋅ Imtiaz Ahmed ⋅ Srinjoy Das

Benchmarking Human-Video-Prompted Robot Adaptation under Deployment Scene Constraints

Xunzhe Zhou ⋅ Yiyang Cai ⋅ Fengyi Wang ⋅ Ran Ju ⋅ Hanxiang Ren ⋅ Ruizhe Liu ⋅ Yu Zhang ⋅ Qian Luo ⋅ Feng Chen ⋅ Pei Zhou ⋅ Yi Ma ⋅ Yanchao Yang

Forward-Only Test-Time Adaptation for Spiking Neural Networks

Ruoyu Zhao ⋅ Yuting Chen ⋅ Jiaqi Wu ⋅ Luziwei Leng

Evaluating Agent Fidelity Beyond Capability for Local Multi-Agent Decisions

Yian Xie ⋅ Delong Li ⋅ Haochen Gong ⋅ Guangsheng Yu ⋅ Baihe Ma ⋅ Xu Wang

Disaggregated Quantization: Specializing LLM Prefill and Decode

Andrei Panferov ⋅ Maximilian Kleinegger ⋅ Sweta Priyadarshi ⋅ Tijmen Blankevoort ⋅ Dan Alistarh

Measured Structure as Self-Supervision: Phase-Equivariant Latent Dynamics for Cyclic Biosignals

Dominic Dootson ⋅ Blaise Delaney

On-Device Financial Claim Verification at Cloud-Level Accuracy

Jayden Chen ⋅ Yan Wang ⋅ Jinnan Yang ⋅ Zhen Bi ⋅ Haihong Yang

Scaling Context Down for On-Device Tabular Foundation Models

Shuting Luo ⋅ Henry Xiao ⋅ Luiz Pizzato ⋅ Ruth A Oliver

Analysis of Input Samples and Internal Representations in Spiking Neural Networks from the Perspective of the Frequency Domain

Yongok Han ⋅ Hyun-Jong Lee ⋅ Jae-Han Lim

Selective On-Device File Routing over User-Defined Taxonomies

Sriram Gutlapalli ⋅ Nihanth Tatikonda ⋅ Apurva Reddy ⋅ Shanaya Pal ⋅ Lashika Kapoor

Depth-Aware Hybrid On-Device/Cloud Perception for Controllable Mobile Photo Editing

Utkarsh Bheemanaboyna

Teachers as Modalities: Shared–Private Factorization for Multi-Teacher Distillation

Tingwang Chen ⋅ Hangke Sui ⋅ Minh Do

The Price of Precision: A Calibrated Cost Model for Mixed-Precision LLM Serving

Mukund Athmaram ⋅ Manoj Rao ⋅ Liang Ma

Quantization of large models with weight outliers and sparsity

Leonidas Tsigkounakis ⋅ Abraham J Mountain ⋅ Yilang Zhang ⋅ Georgios Giannakis

Beyond the Emulator: Bridging the Sim-to-Real Gap in Mobile Agent Benchmarking

Tony Li ⋅ Ramani Tyagi ⋅ Daniel Toyama

Quantizing Pathology Foundation Models for Local Deployment: Clinical Robustness and Backend-Dependent Efficiency

Mohammad Saber Pourheydari ⋅ Chad Vanderbilt ⋅ Christoph Lippert ⋅ Gabriele Campanella

Lightweight, Provenance-Aware Retrieval Gating for Vision–Language Inference on Resource-Constrained Devices

Souparna Chatterjee ⋅ Suchetana Chattopadhyay

Signal-Driven Fusion for Stable Federated Vision-Language Model Personalization

Shahriar Mahir