On-Device Inference

Ask about this topic

Overview

On-device inference studies how to run models locally on phones, wearables, robots, and tiny microcontrollers when connectivity, privacy, latency, memory, or battery life is limited. Common approaches include quantization, pruning and distillation, compact hardware-aware architectures, and runtimes that manage memory, accelerators, and computation across devices. Applications range from keyword spotting and wearable activity recognition to mobile language models, image generation, medical screening, and robot control, evaluated on target hardware and task benchmarks. A central challenge is balancing accuracy and capability against real latency, energy, and memory costs, which vary across hardware; deciding when to compute locally, adapt, or offload remains an open systems trade-off.

Latest papers

CardsList

190 papers18 papers in the last four weeks, down 18% on the four weeks before. 0.1% of all new papers.

May 13, 2026

Bridging the Rural Healthcare Gap: A Cascaded Edge-Cloud Architecture for Automated Retinal Screening

Diabetic Retinopathy (DR) is one of the leading causes of preventable blindness, yet rural regions often lack the specialists and infrastructure needed for early detection. Although cloud-based deep learning systems offer high accuracy, they face significant challenges in these settings due to high latency, limited bandwidth, and high data transmission costs. To address these challenges, we propose a two-tier edge-cloud cascade on the public APTOS 2019 Blindness Detection dataset. Tier 1 runs a lightweight MobileNetV3-small model on a local clinic device to perform a binary triage between Referable DR (Classes 2-4) and Non-referable DR (Classes 0-1). Tier 2 runs a RETFoundDINOv2 model in the cloud for ordinal severity grading, but only on the subset of images flagged as referable by Tier 1. On a stratified APTOS test split of 733 images, Tier 1 reaches 98.99% sensitivity and 84.37% specificity at a validation-tuned high-sensitivity threshold. The default cascade forwards 49.52% of test images to Tier 2, reducing cloud calls by 50.48% relative to using a cloud-based model for all images. In the deployed 4-class output space (Class 0-1 / Class 2 / Class 3 / Class 4), the cascade obtains 80.49% accuracy and 0.8167 quadratic weighted kappa; the cloud-only baseline obtains 80.76% accuracy and 0.8184 quadratic weighted kappa. On APTOS, the cascade cuts cloud use by about half with a modest drop in grading performance. Index Terms: Diabetic Retinopathy, Edge-Cloud Cascade, MobileNetV3-small, RETFound-DINOv2, Retinal Screening, tele-ophthalmology

Nishi Doshi, Shrey Shah

University of Southern California · Los Angeles, USA

Diabetic Retinopathy GradingEdge-Cloud ComputingOn-Device Inference
AskPDF
May 13, 2026

Contextual Bandits for Resource-Constrained Devices using Probabilistic Learning

Contextual bandits (CB) are online sequential decision-making problems under partial feedback that underpin many adaptive services. There is a growing demand to deploy CB agents directly on-device, under strict constraints on memory, compute, and energy. However, standard linear CB algorithms are often impractical for resource-constrained devices with their unfavorable scaling in computational and memory costs. Recently, HD-CB, a CB approach based on hyperdimensional computing principles, has been proposed to model and solve CB problems by moving into high-dimensional spaces. HD-CB offers faster convergence, favorable scalability, and improves memory efficiency compared to linear CB algorithms. However, its learning rule is accumulation-based: the values of action vectors grow over time, requiring high precision. While periodic binarization can prevent overflow in low-precision components, it may discard important information about magnitudes and degrade decision quality. This paper introduces probabilistic HD-CB, a low-precision variant that replaces deterministic accumulation with a probabilistic update rule. At each step, only a random subset of vector components is updated, with a time-decaying update probability, and component values are constrained to a predefined range [-k,+k]. This approach enables low-precision components, prevents overflow without periodic binarization, and reduces the expected update cost in proportion to the fraction of updated components. Off-policy evaluation on standardized synthetic CB benchmarks using the Open Bandit Pipeline shows that probabilistic HD-CB consistently outperforms binarized HD-CB at equal precision, while approaching the performance of HD-CB with as few as 3 bits per component.

Marco Angioli, Kevin Johansson, Antonello Rosato +2

Sapienza University of Rome · Orebro University · Linkoping University +1

Hyperdimensional ComputingContextual BanditsLow-Bit QuantizationOn-Device Inference
AskPDF
May 11, 2026

DECO: Sparse Mixture-of-Experts with Dense-Comparable Performance on End-Side Devices

While Mixture-of-Experts (MoE) scales model capacity without proportionally increasing computation, its massive total parameter footprint creates significant storage and memory-access bottlenecks, which hinder efficient end-side deployment that simultaneously requires high performance, low computational cost, and small storage overhead. To achieve these properties, we present DECO, a sparse MoE architecture designed to match the performance of dense Transformers under identical total parameter budgets and training tokens. DECO utilizes the differentiable and flexible ReLU-based routing enhanced by learnable expert-wise scaling, which adaptively balances the contributions of routed and shared experts. Furthermore, we introduce NormSiLU, an activation function that normalizes inputs prior to SiLU operators, producing a more stable trend of routed-expert activation ratio and a higher intrinsic sparsity level. We also identify an empirical advantage in using non-gated MLP experts with ReLU-based routing, indicating the possibility of MoE architecture simplification. Experiments demonstrate that DECO, activating only 20% of routed experts, matches dense performance and outperforms established MoE baselines. Our specialized acceleration kernel delivers a 2.93×\times speedup on Jetson AGX Orin compared with dense inference. Code and checkpoints are available at https://github.com/thunlp/DECO.

Chenyang Song, Weilin Zhao, Xu Han +3

Dept. of Comp. Sci. & Tech., Institute for AI, Tsinghua University, Beijing, China

Sparse Mixture-of-ExpertsMixture-of-Experts InferenceExpert RoutingOn-Device Inference
AskPDF
May 10, 2026

Split CNN Inference on Networked Microcontrollers

Running deep neural networks on microcontroller units (MCUs) is severely constrained by limited memory resources. While TinyML techniques reduce model size and computation, they often fail in practice due to excessive peak Random Access Memory (RAM) usage during inference, dominated by intermediate activations. As a result, many models remain infeasible on standalone MCUs. In this work, we present a fine-grained split inference system for networked MCUs that enables collaborative inference of Convolutional Neural Networks (CNN) models across multiple devices. Our key insight is that breaking the memory bottleneck requires splitting inference at sub-layer granularity rather than at layer boundaries. We reinterpret pre-trained models to enable kernel-wise and neuron-wise partitioning, and distribute both model parameters and intermediate activations across multiple MCUs. A lightweight, resource-aware coordinator orchestrates the inference across MCU devices with heterogeneous resources. We implement the proposed system on a real testbed and evaluate it on up to 8 MCUs using MobileNetV2, a representative CNN model. Our experimental results show that CNN models infeasible on a single MCU can be executed across networked MCUs, reducing the per-MCU peak RAM usage while maintaining the practical end-to-end inference latency. All the source code of this work can be found here: https://github.com/shashsuresh/split-inference-on-MCUs.

Junyu Lu, Shashwath Suresh, Hao Liu +2

Delft University of Technology, Delft, The Netherlands

Efficient Neural Network InferenceEdge InferenceSplit InferenceConvolutional Neural NetworksOn-Device Inference
AskPDF
May 8, 2026

HARMONY: Bridging the Personalization-Generalization Gap by Mitigating Representation Skew in Heterogeneous Split Federated Learning

Mobile devices face diverse resource constraints and non-IID data class distributions, requiring fast on-device inference for local in-distribution (ID) classes and on-demand remote support for client-specific out-of-distribution (OOD) classes. Hybrid split federated learning (Hybrid SFL) couples personalized client-side front ends (supporting early exit) with a generalized server-side backend for fallback inference, balancing accuracy and cost. However, under client architectural heterogeneity, the existing hybrid SFL suffers from representation skew, where features from customized extractors fail to align in the shared space, leading to a sharp degradation in the server model responsible for OOD prediction. We propose HARMONY, the first hybrid SFL framework to support heterogeneous client architectures. HARMONY modifies meta-learning to simulate diverse extractors across parameters and architectures, and to learn to personalize. To mitigate representation skew, HARMONY conducts server-side contrastive learning to align extracted features, neither sacrificing clients' personalization nor sharing raw labels. Compared to the state of the art across multiple datasets and model families, HARMONY improves test accuracy by up to 43.0%/28.3% without/with OOD, respectively, while maintaining acceptable latency.

Jiseok Youn, You Rim Choi, Goodsol Lee +3

Department of ECE and INMC, Seoul National University, Seoul, South Korea · Graduate School of Data Science, Seoul National University, Seoul, South Korea · Department of Computer Science, University of Colorado Boulder, CO, USA

Meta-LearningOOD GeneralizationSplit Federated LearningRepresentation AlignmentFederated LearningOn-Device Inference+1
AskPDF
May 7, 2026

On-Orbit Real-Time Wildfire Detection Under On-Board Constraints

We present a deployed system for on-orbit wildfire detection aboard a nine-satellite commercial thermal infrared constellation, operating under demanding joint constraints: sub-megabyte model footprint, sub-150 ms per-batch TensorRT FP16 inference on an NVIDIA Jetson Xavier NX, and an end-to-end alert pipeline targeting under 10 minutes from satellite overpass to fire event communication. The system operates on uncalibrated mid-wave infrared (MWIR) single-band imagery at 200 m ground sampling distance, where fires frequently appear as sub-pixel or single-pixel thermal anomalies under extreme class imbalance -- challenges not addressed by the contextual thermal-thresholding pipelines (MODIS, VIIRS) that currently dominate operational fire monitoring. We present an empirical study of lightweight dense representation learning for this regime using a proprietary nine-satellite MWIR dataset. We compare dense masked autoencoding (DenseMAE) and a hybrid DenseMAE+EMA (exponential moving average) distillation variant, and evaluate representations via linear probing and full-distribution pixel-level average precision (AP) under extreme class imbalance. DenseMAE pretraining enables compact downstream models on the latency-accuracy Pareto frontier: our fastest SSL-pretrained model achieves 0.640 test AP and 0.69 event-level Fire-F1 with 65.34 ms latency per batch and a 0.52 MB engine, without pruning or compression. The best configuration reaches 0.699 AP and 0.744 Fire-F1 below 1 MB, outperforming a supervised baseline (0.650 AP) under comparable constraints.

Matthias Rötzer, Veronika Pörtge, Martin Ickerott +6

OroraTech GmbH, St.-Martin-Str. 112, 81669 Munich, Germany · Ludwig-Maximilians-Universität München, Munich, Germany

Efficient Neural Network InferenceWildfire DetectionInfrared Small Target DetectionRemote SensingOn-Device InferenceMasked Image Modeling
AskPDF
May 7, 2026

PoTAcc: A Pipeline for End-to-End Acceleration of Power-of-Two Quantized DNNs

Power-of-two (PoT) quantization significantly reduces the size of deep neural networks (DNNs) and replaces multiplications with bit-shift operations for inference. Prior work has shown that PoT-quantized DNNs can preserve accuracy for tasks such as image classification; however, their performance on resource-constrained edge devices remains insufficiently understood. While general-purpose edge CPUs and GPUs do not provide optimized backends for bit-shift operations, custom hardware accelerators can better exploit PoT quantization by implementing dedicated shift-based processing elements. However, deploying PoT-quantized models on such accelerators is challenging due to limited support in existing inference frameworks. In addition, the impact of different PoT quantization strategies on hardware design, performance, and energy efficiency during full inference has not been systematically explored. To address these challenges, we propose PoTAcc, an open-source end-to-end pipeline for accelerating and evaluating PoT-quantized DNNs on resource-constrained edge devices. PoTAcc enables seamless preparation and deployment of PoT-quantized models via TensorFlow Lite (TFLite) across heterogeneous platforms, including CPU-only systems and hybrid CPU-FPGA systems with custom accelerators. We design shift-based processing element (shift-PE) accelerators for three PoT quantization methods and implement them on two FPGA platforms. We evaluate accuracy, performance, energy efficiency, and resource utilization across a range of models, including CNNs and Transformer-based architectures. Results show that our CPU-accelerator design achieves up to 3.6x speedup and 78% energy reduction compared to CPU-only execution for PoT-quantized DNNs on PYNQ-Z2 and Kria boards. The code will be publicly released at https://github.com/gicLAB/PoTAcc

Rappy Saha, Jude Haris, Nicolas Bohm Agostini +2

Efficient Neural Network InferenceFPGA-Based Neural Network AccelerationNeural Network QuantizationOn-Device Inference
AskPDF
May 5, 2026

ExecuTorch -- A Unified PyTorch Solution to Run AI Models On-Device

Local execution of AI on edge devices is important for low latency and offline operation. However, deploying models on diverse hardware remains fragmented, often requiring model conversion or complete reimplementation outside the PyTorch ecosystem where the model was originally authored. We introduce ExecuTorch, a unified PyTorch-native deployment framework for edge AI. ExecuTorch enables seamless deployment of machine learning models across heterogeneous compute environments. It scales from embedded microcontrollers to complex system-on-chips (SoCs) with dedicated accelerators, powering devices ranging from wearables and smartphones to large compute clusters. ExecuTorch preserves PyTorch semantics while allowing customization, support for optimizations like quantization, and pluggable execution "backends". These features together enable fast experimentation, allowing researchers to validate deployment behavior entirely within PyTorch, bridging the gap between research and production.

Mergen Nachin, Digant Desai, Sicheng Stephen Jia +36

Meta · Work done while at Meta.

Heterogeneous ComputingEdge MLOn-Device InferenceEdge AI
AskPDF
May 5, 2026

QKVShare: Quantized KV-Cache Handoff for Multi-Agent On-Device LLMs

Multi-agent LLM systems on edge devices need to hand off latent context efficiently, but the practical choices today are expensive re-prefill or full-precision KV transfer. We study QKVShare, a framework for quantized KV-cache handoff between agents that combines token-level mixed-precision allocation, a self-contained CacheCard representation, and a HuggingFace-compatible cache injection path. Our current results support a narrower but clearer story than the original draft: on 150 GSM8K problems with Llama-3.1-8B-Instruct, adaptive quantization remains competitive under repeated handoff and shows its clearest gains against uniform quantization in deeper-hop, higher budget settings; for handoff latency, the QKVShare path reduces TTFT relative to full re prefill at every tested context, from 130.7 ms vs. 150.2 ms at nominal 1K context to 397.1 ms vs. 1029.7 ms at nominal 8K context;. Stage timing shows that post-injection generation, not card creation, dominates the current QKVShare latency path. These results position quantized KV handoff as a promising on-device systems direction while also highlighting the need for stronger controller ablations and apples-to-apples runtime comparisons.

Pratik Honavar, Tejpratap GVSL

Mixed-Precision QuantizationMulti-Agent LLM SystemsKV CachingKV-Cache ManagementKV-Cache QuantizationOn-Device Inference
AskPDF
May 5, 2026

Real Image Denoising with Knowledge Distillation for High-Performance Mobile NPUs

While deep-learning-based image restoration has achieved unprecedented fidelity, deployment on mobile Neural Processing Units (NPUs) remains bottlenecked by operator incompatibility and memory-access overhead. We propose an NPU-aware hardware-algorithm co-design approach for real-world image denoising on mobile NPUs. Our approach employs a high-capacity teacher to supervise a lightweight student network specifically designed to leverage the tiled-memory architectures of modern mobile SoCs. By prioritizing NPU-native primitives -- standard 3x3 convolutions, ReLU activations, and nearest-neighbor upsampling -- and employing a progressive context expansion strategy (up to 1024x1024 crops), the model achieves 37.66 dB PSNR / 0.9278 SSIM on the validation benchmark and 37.58 dB PSNR / 0.9098 SSIM on the held-out test benchmark at full resolution (2432x3200) in the Mobile AI 2026 challenge. Following the official challenge rules, the inference runtime is measured under a standardized Full HD (1088x1920) protocol, where it runs in 34.0 ms on the MediaTek Dimensity 9500 and 46.1 ms on the Qualcomm Snapdragon 8 Elite NPU. We further reveal an "Inference Inversion" effect, where strict adherence to NPU-compatible operations enables dedicated NPU execution up to 3.88x faster than the integrated mobile GPU. The 1.96M-parameter student recovers 99.8% of the teacher's restoration quality via high-alpha knowledge distillation (alpha = 0.9), achieving a 21.2x parameter reduction while closing the PSNR gap from 1.63 dB to only 0.05 dB. These results establish hardware-aware distillation as an effective strategy for unifying high-fidelity denoising with practical deployment across diverse mobile NPU architectures. The proposed lightweight student model (LiteDenoiseNet) and its training statistics are provided in the NN Dataset, available at https://github.com/ABrain-One/NN-Dataset.

Faraz Kayani, Sarmad Kayani, Asad Ahmed +2

Computer Vision Lab, CAIDAS & IFI, University of Würzburg, Germany

Efficient Neural Network InferenceKnowledge DistillationOn-Device InferenceImage Denoising
AskPDF
May 4, 2026

Mixed-Precision Information Bottlenecks for On-Device Trait-State Disentanglement in Bipolar Agitation Detection

Continuous monitoring of bipolar disorder agitation via voice biomarkers requires disentangling stable speaker traits from volatile affective states on resource-constrained edge devices. We introduce MP-IB, the first framework to treat mixed-precision quantization as an information bottleneck for clinical trait-state separation. The core insight is that numerical precision itself controls capacity: an FP16 trait head (1,024 bits) encodes speaker identity, while an INT4 state head (128 bits) captures agitation, yielding 8x information asymmetry without adversarial training. We augment this with Dynamic Precision Scheduling and Multi-Scale Temporal Fusion. On Bridge2AI-Voice (N=833, 4 sessions/participant, strict speaker-independent CV), MP-IB achieves rho = 0.117 (95% CI: [0.089, 0.145], p=0.003 vs. chance), outperforming 94M-parameter WavLM-Adapter with in-domain SSL continuation (rho = -0.042), beta VAE disentanglement (rho = 0.089), and hand-crafted prosody (rho = 0.031) by 2.8--15.9 points absolute. Zero-shot transfer to CREMA-D achieves AUC=0.817. Identity leakage is suppressed to near-random (EER=0.42, MIA-AUC=0.52). End-to-end latency is 23.4 ms with a 617 KB footprint, enabling real-time monitoring on sub 20 dollar devices.

Joydeep Chandra

Department of Computer Science & Technology, Tsinghua University, Beijing, China.

Mixed-Precision QuantizationInformation BottleneckAudio UnderstandingSpeaker DisentanglementOn-Device Inference
AskPDF
May 2, 2026

AgriKD: Cross-Architecture Knowledge Distillation for Efficient Leaf Disease Classification

Automated leaf disease classification is critical for early disease detection in resource-constrained field environments. Vision Transformers (ViTs) provide strong representation capability by modeling long-range dependencies and inter-class relationships; however, their high computational cost makes them impractical for deployment on edge devices. As a result, existing approaches struggle to effectively transfer these rich representations to lightweight models. This paper introduces AgriKD, a cross-architecture knowledge distillation framework for efficient edge deployment, which transfers knowledge from a Vision Transformer (ViT) teacher to a compact convolutional student model. To bridge the representational gap between Transformer and CNN architectures, the proposed approach integrates multiple distillation objectives at the output, feature, and relational levels, where each objective captures a different aspect of the teacher knowledge. This enables the student model to better preserve and utilize transformer-derived global representations. Experiments on multiple leaf disease datasets show that the distilled student achieves performance comparable to the teacher while significantly improving efficiency, reducing model parameters by approximately 172 times, computational cost by 47.57 times, and inference latency by 18-22 times. Furthermore, the optimized model is deployed across multiple runtime formats, including ONNX, TFLite Float16, and TensorRT FP16, achieving consistent predictive performance with negligible accuracy degradation. Real-world deployment on NVIDIA Jetson edge devices and a mobile application demonstrates reliable real-time inference, highlighting the practicality of AgriKD for AI-powered agricultural applications in resource-constrained environments.

Minh-Dung Le, Minh-Duc Hoang, Hoang-Vu Truong +1

AIT laboratory, Faculty of Artificial Intelligence, FPT University, Da Nang, 550000, Viet Nam

Efficient Neural Network InferencePlant Disease ClassificationCross-Architecture Knowledge DistillationKnowledge DistillationOn-Device Inference
AskPDF
Apr 29, 2026

Efficient, VRAM-Constrained xLM Inference on Clients

To usher in the next round of client AI innovation, there is an urgent need to enable efficient, lossless inference of high-accuracy large language models (LLMs) and vision language models (VLMs), jointly referred to as xLMs, on client systems. To address this, we present pipelined sharding, a novel, benchmark-profile-guided CPU-GPU hybrid scheduling technique to achieve efficient, VRAM-constrained inference for both dense and mixture-of-experts (MoE) LLMs. Using a combination of model sharding at the sub-layer level, CPU offloading, pipelined copy-compute, and prioritized tensor placement in VRAM, it optimizes both time-to-first-token (TTFT) and tokens per second (TPS) metrics, while flexibly adapting to system and inference conditions. For efficient, high-accuracy VLM inference, we combine pipelined sharding with a llama..cpp implementation of three well-understood prior ideas (jointly called VLMOpt), namely, vision tensor CPU offloading, flash attention, and vision and language model VRAM overlap avoidance. These enhancements are targeted at improving client xLM inference in future releases of two important NVIDIA products - the In-Game Inferencing software development kit (IGI SDK) and the Cosmos-Reason1 (CR1) physical AI reasoning VLM. Highlights from our rigorous evaluation spanning multiple models and client systems include: for interactive use, TTFT improves by up to 6.7x and TPS by up to 30x for LLMs, and CR1 inference's VRAM demand is down by 10x, while in batched mode, throughput improves by up to 8.2x, all compared to their respective aggressive baselines. This paper is accepted at the 9th MLSys Conference (Industry Track), 2026. Code and artifact available at: https://github.com/deepshnv/pipeshard-mlsys26-ae

Aditya Ukarande, Deep Shekhar, Marc Blackstein +1

NVIDIA Corporation, Santa Clara, CA, USA · NVIDIA Graphics Pvt Ltd., Bangalore, India

GPU AccelerationEfficient VLM InferenceLLM Inference SchedulingMemory-Efficient InferenceLLM Inference AccelerationPipeline Parallelism+1
AskPDF
Apr 28, 2026

At the Edge of the Heart: ULP FPGA-Based CNN for On-Device Cardiac Feature Extraction in Smart Health Sensors for Astronauts

The convergence of accelerating human spaceflight ambitions and critical terrestrial health monitoring demands is driving unprecedented requirements for reliable, real-time feature extraction on extremely resource-constrained wearable health sensors. We present an ultra-low-power (ULP) Field-Programmable Gate Array (FPGA) based solution for real-time Seismocardiography (SCG) feature classification using Convolutional Neural Networks (CNNs). Our approach combines quantization-aware training with a systolic-array accelerator to enable efficient integer-only inference on the Lattice iCE40UP5K FPGA, which offers an ideal platform for battery-powered deployments -- particularly in space environments -- thanks to its power efficiency and radiation resilience. The implementation achieves a validation accuracy of 98% while consuming only 8.55 mW, completing inference in 95.5 ms with minimal hardware resources (2,861 LUTs and 7 DSP blocks). These results demonstrate that fully on-device SCG-based cardiac feature extraction is feasible on resource-constrained hardware, enabling energy-efficient, autonomous health monitoring for astronauts in long-duration space missions.

Kazi Mohammad Abidur Rahman, Davis Rakhshan, Philipp Lütke +2

Smart Sensors Group, Hamburg University of Technology, Hamburg, Germany · Distributed Systems, Kiel University, Kiel, Germany · Networked Cyber-Physical Systems, Hamburg University of Technology, Hamburg, Germany

Time Series ClassificationWearable Health MonitoringFPGA-Based Neural Network AccelerationOn-Device Inference
AskPDF
Apr 28, 2026

PI-TTA: Physics-Informed Source-Free Test-Time Adaptation for Robust Human Activity Recognition on Mobile Devices

Source-free test-time adaptation (TTA) is appealing for mobile and wearable sensing because it enables on-device personalization from unlabeled test streams without centralizing private data. However, sensor-based human activity recognition (HAR) poses challenges that are less pronounced in standard vision benchmarks: behavioral inertial streams are temporally correlated and often exhibit within-session shifts caused by sensor rotation, placement change, and sampling-rate drift. Under this streaming non-i.i.d. setting, widely used vision-style TTA objectives can become unstable, leading to overconfident errors, representation collapse, and catastrophic forgetting. We propose PI-TTA, a lightweight source-free adaptation framework that stabilizes online updates through three physics-consistent constraints: gravity consistency, short-horizon temporal continuity, and spectral stability. PI-TTA updates the same small parameter subset as strong source-free baselines and incurs only modest overhead, making it suitable for on-device deployment. Experiments on USCHAD, PAMAP2, and mHealth under long-sequence stress tests and factorized shift protocols show that PI-TTA mitigates the severe degradation observed in confidence-driven baselines and preserves stable adaptation under sustained streaming conditions. It improves long-sequence accuracy by up to 9.13% and reduces physical-violation rates by 27.5%, 24.1%, and 45.4% on USCHAD, PAMAP2, and mHealth, respectively. These results demonstrate that physics-informed adaptation can improve accuracy, stability, and deployment reliability for real-world mobile sensing systems.

Changyu Li, Lu Wang, Ming Lei +4

School of Computing and Information Technology, Great Bay University, Dongguan, China · College of Computer Science and Software Engineering, Shenzhen University, Shenzhen, China · University of Warwick, Coventry, U.K. +2

Test-Time AdaptationHuman Activity RecognitionWearable SensingPhysics-Informed MLOn-Device Inference
AskPDF
Apr 27, 2026

Characterizing Vision-Language-Action Models across XPUs: Constraints and Acceleration for On-Robot Deployment

Vision-Language-Action (VLA) models are promising for generalist robot control, but on-robot deployment is bottlenecked by real-time inference under tight cost and energy budgets. Most prior evaluations rely on desktop-grade GPUs, obscuring the trade-offs and opportunities offered by heterogeneous edge accelerators (GPUs/XPUs/NPUs). We present a systematic analysis for low-cost VLA deployment via model-hardware co-characterization. First, we build a cross-accelerator leaderboard and evaluate model-hardware pairs under CET (Cost, Energy, Time), showing that right-sized edge devices can be more cost-/energy-efficient than flagship GPUs while meeting control-rate constraints. Second, using in-depth profiling, we uncover a consistent two-phase inference pattern: a compute-bound VLM backbone followed by a memory-bound Action Expert, which induces phase-dependent underutilization and hardware inefficiency. Finally, guided by these insights, we propose DP-Cache and V-AEFusion to reduce diffusion redundancy and enable asynchronous pipeline parallelism, achieving up to 2.9x speedup on GPUs and 6x on edge NPUs with only marginal success degradation. The example leaderboard website is available at: https://vla-leaderboard-01.vercel.app/.

Kaijun Zhou, Qiwei Chen, Da Peng +3

School of Computer Science, Shanghai Jiao Tong University

AI Accelerator InferenceEdge InferenceEfficient VLA Model InferenceVision-Language-Action ModelsVLMs for RoboticsDiffusion Model Inference Acceleration+1
AskPDF
Apr 27, 2026

GoClick: Lightweight Element Grounding Model for Autonomous GUI Interaction

Graphical User Interface (GUI) element grounding (precisely locating elements on screenshots based on natural language instructions) is fundamental for agents interacting with GUIs. Deploying this capability directly on resource-constrained devices like mobile phones is increasingly critical for GUI agents requiring low latency. However, this goal faces a significant challenge, as current visual grounding methods typically employ large vision-language model (VLM) (more than 2.5B parameters), making them impractical for on-device execution due to memory and computational constraints. To address this, this paper introduces GoClick, a lightweight GUI element grounding VLM with only 230M parameters that achieves excellent visual grounding accuracy, even on par with significantly larger models. Simply downsizing existing decoder-only VLMs is a straightforward way to design a lightweight model, but our experiments reveal that this approach yields suboptimal results. Instead, we select an encoder-decoder architecture, which outperforms decoder-only alternatives at small parameter scales for GUI grounding tasks. Additionally, the limited capacity of small VLMs encourages us to develop a Progressive Data Refinement pipeline that utilizes task type filtering and data ratio adjustment to extract a high-quality 3.8M-sample core set from a 10.8M raw dataset. Training GoClick using this core set brings notable grounding accuracy gains. Our experiments show that GoClick excels on multiple GUI element grounding benchmarks while maintaining a small size and high inference speed. GoClick also enhances GUI agent performance when integrated into a device-cloud collaboration framework, where GoClick helps cloud-based task planners perform precise element localization and achieve higher success rates. We hope our method serves as a meaningful exploration within the GUI agent community.

Hongxin Li, Yuntao Chen, Zhaoxiang Zhang

University of Chinese Academy of Sciences, Beijing, China. · New Laboratory of Pattern Recognition, State Key Laboratory of Multimodal Artificial Intelligence Systems, Institute of Automation, Chinese Academy of Sciences, Beijing, China. · Hong Kong Institute of Science & Innovation, Chinese Academy of Sciences, Hong Kong, China.

Efficient VLM InferenceTraining Data CurationGUI AgentsGUI GroundingOn-Device Inference
AskPDF
Apr 24, 2026

On-Device Vision Training, Deployment, and Inference on a Thumb-Sized Microcontroller

This paper presents a complete, end-to-end on-device vision machine learning pipeline, comprising data acquisition, two-layer CNN training with Adam optimization, and real-time inference, executing entirely on a microcontroller-class device costing $15-40 USD. Unlike cloud-based workflows that require external infrastructure and conceal the computational pipeline from the practitioner, this system implements every step of the core ML lifecycle in approximately 1,750 lines of readable C++ that compiles in under one minute using the Arduino IDE, with no external ML dependencies. Running on the Seeed Studio ESP32-S3 XIAO ML Kit (8 MB PSRAM), the firmware achieves three-class 64x64 image classification in approximately 9 minutes per training run, with real-time inference at 6.3 FPS. Key contributions include: correct batch-level gradient accumulation; pre-computed resize lookup tables for inference; dual-format weight export for SD-free baked-in deployment; a three-tier weight priority system (SD binary > baked-in header > He-initialization) resolved automatically at boot; a single-constant network reconfiguration interface; and PSRAM-aware memory management suited to microcontroller constraints. All source code and reference datasets are released under the MIT License at https://github.com/webmcu-ai/on-device-vision-ai

Jeremy Ellis

High School Robotics Educator, British Columbia, Canada

On-Device TrainingTiny MLEdge MLImage ClassificationOn-Device Inference
AskPDF
Apr 23, 2026

Bridging the Training-Deployment Gap: Gated Encoding and Multi-Scale Refinement for Efficient Quantization-Aware Image Enhancement

Image enhancement models for mobile devices often struggle to balance high output quality with the fast processing speeds required by mobile hardware. While recent deep learning models can enhance low-quality mobile photos into high-quality images, their performance is often degraded when converted to lower-precision formats for actual use on mobile phones. To address this training-deployment mismatch, we propose an efficient image enhancement model designed specifically for mobile deployment. Our approach uses a hierarchical network architecture with gated encoder blocks and multiscale refinement to preserve fine-grained visual features. Moreover, we incorporate Quantization-Aware Training (QAT) to simulate the effects of low-precision representation during the training process. This allows the network to adapt and prevents the typical drop in quality seen with standard post-training quantization (PTQ). Experimental results demonstrate that the proposed method produces high-fidelity visual output while maintaining the low computational overhead needed for practical use on standard mobile devices. The code will be available at https://github.com/GenAI4E/QATIE.git.

Dat To-Thanh, Nghia Nguyen-Trong, Hoang Vo +2

University of Science, VNU-HCM, Vietnam · Vietnam National University, Ho Chi Minh City, Vietnam · University of Information Technology, VNU-HCM, Vietnam +2

Image EnhancementOn-Device Inference
AskPDF
Apr 21, 2026

KD-Judge: A Knowledge-Driven Automated Judge Framework for Functional Fitness Movements on Edge Devices

Functional fitness movements are widely used in training, competition, and health-oriented exercise programs, yet consistently enforcing repetition (rep) standards remains challenging due to subjective human judgment, time constraints, and evolving rules. Existing AI-based approaches mainly rely on learned scoring or reference-based comparisons and lack explicit rule-based, limiting transparency and deterministic rep-level validation. To address these limitations, we propose KD-Judge, a novel knowledge-driven automated judging framework for functional fitness movements. It converts unstructured rulebook standards into executable, machine-readable representations using an LLM-based retrieval-augmented generation and chain-of-thought rule-structuring pipeline. The structured rules are then incorporated by a deterministic rule-based judging system with pose-guided kinematic reasoning to assess rep validity and temporal boundaries. To improve efficiency on edge devices, including a high-performance desktop and the resource-constrained Jetson AGX Xavier, we introduce a dual strategy caching mechanism that can be selectively applied to reduce redundant and unnecessary computation. Experiments demonstrate reliable rule-structuring performance and accurate rep-level assessment, with judgment evaluation conducted on the CFRep dataset, achieving faster-than-real-time execution (real-time factor (RTF) < 1). When the proposed caching strategy is enabled, the system achieves up to 3.36x and 15.91x speedups on resource-constrained edge device compared to the non-caching baseline for pre-recorded and live-streaming scenarios, respectively. These results show that KD-Judge enables transparent, efficient, and scalable rule-grounded rep-level analysis that can complement human judging in practice.

Shaibal Saha, Fan Li, Yunge Li +3

Department of Computer Science and Engineering, Oakland University, Rochester Hills, USA · Intelligent Data Management and Analytics LLC, USA

On-Device Inference
AskPDF
Apr 20, 2026

Efficient Mixture-of-Experts LLM Inference with Apple Silicon NPUs

Apple Neural Engine (ANE) is a dedicated neural processing unit (NPU) present in every Apple Silicon chip. Mixture-of-Experts (MoE) LLMs improve inference efficiency via sparse activation but are challenging for NPUs in three ways: expert routing is unpredictable and introduces dynamic tensor shapes that conflict with the shape-specific constraints of NPUs; several irregular operators, e.g., top-k, scatter/gather, etc., are not NPU-friendly; and launching many small expert kernels incurs substantial dispatch and synchronization overhead. NPUs are designed to offload AI compute from CPU and GPU; our goal is to enable such offloading for MoE inference, particularly during prefill, where long-context workloads consume substantial system resources. This paper presents NPUMoE, a runtime inference engine that accelerates MoE execution on Apple Silicon by offloading dense, static computation to NPU, while preserving a CPU/GPU fallback path for dynamic operations. NPUMoE uses offline calibration to estimate expert capacity and popularity that drives three key techniques: (1) Static tiers for expert capacity to address dynamic expert routing (2) Grouped expert execution to mitigate NPU concurrency limits (3) Load-aware expert compute graph residency to reduce CPU-NPU synchronization overhead. Experiments on Apple M-series devices using three representative MoE LLMs and four long-context workloads show that NPUMoE consistently outperforms baselines, reducing latency by 1.32x-5.55x, improving energy efficiency by 1.81x-7.37x, and reducing CPU-cycle usage by 1.78x-5.54x through effective NPU offloading.

Afsara Benazir, Felix Xiaozhu Lin

University of Virginia Charlottesville, Virginia, USA

AI Accelerator InferenceEnergy-Efficient MLMixture-of-Experts InferenceLLM Inference AccelerationOn-Device Inference
AskPDF
Apr 20, 2026

Towards Real-Time ECG and EMG Modeling on μμNPUs

The miniaturisation of neural processing units (NPUs) and other low-power accelerators has enabled their integration into microcontroller-scale wearable hardware, supporting near-real-time, offline, and privacy-preserving inference. Yet physiological signal analysis has remained infeasible on such hardware; recent Transformer-based models show state-of-the-art performance but are prohibitively large for resource- and power-constrained hardware and incompatible with μμNPUs due to their dynamic attention operations. We introduce PhysioLite, a lightweight, NPU-compatible model architecture and training framework for ECG/EMG signal analysis. Using learnable wavelet filter banks, CPU-offloaded positional encoding, and hardware-aware layer design, PhysioLite reaches performance comparable to state-of-the-art Transformer-based foundation models on ECG and EMG benchmarks, while being <10% of the size (∼\sim370KB with 8-bit quantization). We also profile its component-wise latency and resource consumption on both the MAX78000 and HX6538 WE2 μμNPUs, demonstrating its viability for signal analysis on constrained, battery-powered hardware. We release our model(s) and training framework at: https://github.com/j0shmillar/physiolite.

Josh Millar, Ashok Samraj Thangarajan, Soumyajit Chatterjee +1

Imperial College London London, United Kingdom · Nokia Bell Labs London, United Kingdom · Brave Software & University of Cambridge London, United Kingdom

Efficient Neural Network InferenceElectrocardiographyOn-Device Inference
AskPDF
Apr 19, 2026

The First Challenge on Mobile Real-World Image Super-Resolution at NTIRE 2026: Benchmark Results and Method Overview

This paper provides a review of the NTIRE 2026 challenge on mobile real-world image super-resolution, highlighting the proposed solutions and the resulting outcomes. The challenge aims to recover high-resolution (HR) images from low-resolution (LR) counterparts generated through unknown degradations with a x4 scaling factor while ensuring the models remain executable on mobile devices. The objective is to develop effective and efficient network designs or solutions that achieve state-of-the-art real-world image super-resolution performance. The track of the challenge evaluates performance using a weighted combination of image quality assessment (IQA) score and speedup ratios. The competition attracted 108 registrants, with 16 teams achieving a valid score in the final ranking. This collaborative effort advances the performance of mobile real-world image super-resolution while offering an in-depth overview of the latest trends in the field.

Jiatong Li, Zheng Chen, Kai Liu +91

Image Super-ResolutionOn-Device Inference
AskPDF
Apr 18, 2026

MobileAgeNet: Lightweight Facial Age Estimation for Mobile Deployment

Mobile deployment of facial age estimation requires models that balance predictive accuracy with low latency and compact size. In this work, we present MobileAgeNet, a lightweight age-regression framework that achieves an MAE of 4.65 years on the UTKFace held-out test set while maintaining efficient on-device inference with an average latency of 14.4 ms measured using the AI Benchmark application. The model is built on a pretrained MobileNetV3-Large backbone combined with a compact regression head, enabling real-time prediction on mobile devices. The training and evaluation pipeline is integrated into the NN LEMUR Dataset framework, supporting reproducible experimentation, structured hyperparameter optimization, and consistent evaluation. We employ bounded age regression together with a two-stage fine-tuning strategy to improve training stability and generalization. Experimental results show that MobileAgeNet achieves competitive accuracy with 3.23M parameters, and that the deployment pipeline from PyTorch training through ONNX export to TensorFlow Lite conversion - preserves predictive behavior without measurable degradation under practical on-device conditions. Overall, this work provides a practical, deployment-ready baseline for mobile-oriented facial age estimation.

Arun Kumar, Aswathy Baiju, Radu Timofte +1

Computer Vision Lab, CAIDAS & IFI, University of Würzburg, Germany

On-Device Inference
AskPDF
Apr 17, 2026

AdaVFM: Adaptive Vision Foundation Models for Edge Intelligence via LLM-Guided Execution

Language-aligned vision foundation models (VFMs) enable versatile visual understanding for always-on contextual AI, but their deployment on edge devices is hindered by strict latency and power constraints. We present AdaVFM, an adaptive framework for efficient on-device inference of language-aligned VFMs that dynamically adjusts computation based on scene context and task complexity. Our key insight is that the effect of model size reduction on performance is task-dependent in vision applications, motivating a runtime-adaptive execution strategy. AdaVFM integrates neural architecture search (NAS) into the language-aligned VFM backbone to enable lightweight subnet execution during runtime. A multimodal large language model (LLM) deployed on the cloud enables runtime control with a context-aware agent. This synergy allows efficient model adaptation under diverse conditions while maintaining strong accuracy. Extensive experiments on zero-shot classification and open-vocabulary segmentation demonstrate that AdaVFM achieves state-of-the-art accuracy-efficiency trade-offs, surpassing prior baselines by up to 7.9%7.9\% in acc@1 on IN1K and 5.2%5.2\% mIoU on ADE20K over the best models of comparable VFM sizes. For models with similar accuracy, AdaVFM further reduces average FLOPs by up to 77.9%77.9\%.

Yiwei Zhao, Yi Zheng, Huapeng Su +9

Carnegie Mellon University · Meta

Vision Foundation ModelsVision Foundation Model AdaptationOn-Device Inference
AskPDF
Apr 16, 2026

Pushing the Limits of On-Device Streaming ASR: A Compact, High-Accuracy English Model for Low-Latency Inference

Deploying high-quality automatic speech recognition (ASR) on edge devices requires models that jointly optimize accuracy, latency, and memory footprint while operating entirely on CPU without GPU acceleration. We conduct a systematic empirical study of state-of-the-art ASR architectures, encompassing encoder-decoder, transducer, and LLM-based paradigms, evaluated across batch, chunked, and streaming inference modes. Through a comprehensive benchmark of over 50 configurations spanning OpenAI Whisper, NVIDIA Nemotron, Parakeet TDT, Canary, Conformer Transducer, and Qwen3-ASR, we identify NVIDIA's Nemotron Speech Streaming as the strongest candidate for real-time English streaming on resource-constrained hardware. We then re-implement the complete streaming inference pipeline in ONNX Runtime and conduct a controlled evaluation of multiple post-training quantization strategies, including importance-weighted k-quant, mixed-precision schemes, and round-to-nearest quantization, combined with graph-level operator fusion. These optimizations reduce the model from 2.47 GB to as little as 0.67 GB while maintaining word error rate (WER) within 1% absolute of the full-precision PyTorch baseline. Our recommended configuration, the int4 k-quant variant, achieves 8.20% average streaming WER across eight standard benchmarks, running comfortably faster than real-time on CPU with 0.56 s algorithmic latency, establishing a new quality-efficiency Pareto point for on-device streaming ASR.

Nenad Banfic, David Fan, Kunal Vaishnavi +5

CoreAI, Microsoft

Efficient Neural Network InferenceStreaming ASRPost-Training QuantizationOn-Device Inference
AskPDF
Apr 12, 2026

LAWS: Learning from Actual Workloads Symbolically -- A Self-Certifying Parametrized Cache Architecture for Neural Inference, Robotics, and Edge Deployment

We introduce LAWS (Learning from Actual Workloads Symbolically), a self-certifying inference caching architecture that builds a growing library of certified expert functions from deployment observations. Each expert covers a region of input space defined by a node in the Probabilistic Language Trie (PLT) of the base model and carries a formal error bound holding uniformly over all inputs. The central result is a self-certification theorem: for any input x, the LAWS approximation error is bounded by epsilon_fit + 2*Lambda(W)*C_E, where Lambda(W) is the model Lipschitz constant, C_E is the maximum embedding diameter, and epsilon_fit is the expert training error -- all checkable at deployment time without ground truth. We prove that LAWS generalizes both Mixture-of-Experts and KV prefix caching as special cases and is strictly more expressive than any fixed-K MoE or finite cache. Further results include a monotone hit rate theorem (any-match routing ensures coverage only increases), an expert library growth rate of O(2^H log N) where H is workload entropy, a fleet learning convergence theorem with Omega(K) speedup for K-unit fleets, and an over-the-air update bandwidth bound. We conjecture that LAWS is acquisition-optimal among stationary online caching algorithms and that the effective Lipschitz constant on the training distribution grows polynomially rather than exponentially in depth. Applications are developed for LLM inference, robotic control, and multi-agent edge deployment.

Gregory Magarshak

Efficient InferenceLLM Inference AccelerationOn-Device Inference
AskPDF
Apr 11, 2026

FastSHADE: Fast Self-augmented Hierarchical Asymmetric Denoising for Efficient inference on mobile devices

Real-time image denoising is essential for modern mobile photography but remains challenging due to the strict latency and power constraints of edge devices. This paper presents FastSHADE (Fast Self-augmented Hierarchical Asymmetric Denoising), a lightweight U-Net-style network tailored for real-time, high-fidelity restoration on mobile GPUs. Our method features a multi-stage architecture incorporating a novel Asymmetric Frequency Denoising Block (AFDB) that decouples spatial structure extraction from high-frequency noise suppression to maximize efficiency, and a Spatially Gated Upsampler (SGU) that optimizes high-resolution skip connection fusion. To address generalization, we introduce an efficient Noise Shifting Self-Augmentation strategy that enhances data diversity without inducing domain shifts. Evaluations on the MAI2021 benchmark demonstrate that our scalable model family establishes a highly efficient speed-fidelity trade-off. Our base FastSHADE-M variant maintains real-time latency (<50 ms on an Adreno 840 GPU) while preserving structural integrity, and our scaled-up FastSHADE-XL establishes a new state-of-the-art for overall image quality, achieving 37.94 dB PSNR.

Nikolay Falaleev

Fanis London, UK

Image DenoisingOn-Device InferenceU-Net
AskPDF
Apr 11, 2026

Taming Asynchronous CPU-GPU Coupling for Frequency-aware Latency Estimation on Mobile Edge

Precise estimation of model inference latency is crucial for time-critical mobile edge applications, enabling devices to calculate latency margins against deadlines and trade them for enhanced model performance or resource savings. However, the ubiquity of Dynamic Voltage and Frequency Scaling (DVFS) renders traditional static profiling invalid in real-world deployments, as inference latency fluctuates with varying processor (CPU and GPU) frequencies. While extensive profiling across frequency combinations is theoretically possible, it is prohibitively expensive, particularly for emerging Small Language Models (SLMs), where variable context lengths explode the profiling up to days. We observe that simple analytic scaling fails to predict these fluctuations due to the complex asynchronous coupling between CPU (kernel launching) and GPU (execution). In this paper, we introduce FLAME to accurately estimate inference latency across frequency combinations. It features a novel layer-wise modeling that quantifies the overlapping parallelism and then aggregates dynamic pipeline bubbles caused by asynchronous processor interactions when extending to the full model. This bottom-up approach ensures generalizability across diverse models from DNNs to SLMs, and its precise modeling allows for profiling a sparse subset of samples, cutting DNN profiling from hours to minutes and SLM profiling from days to mere minutes, while maintaining small estimation errors across frequencies. We further showcase FLAME's utility in a deadline-aware DVFS, outperforming the state-of-the-art approach in both power efficiency and latency guarantees.

Jiesong Chen, Jun You, Zhidan Liu +1

Department of Computer Science, City University of Hong Kong, Hong Kong, China · INTR Thrust, System Hub, The Hong Kong University of Science and Technology (Guangzhou), Guangzhou, China

Edge InferenceEfficient Neural Network InferenceOn-Device Inference
AskPDF
Apr 10, 2026

DeFakeQ: Enabling Real-Time Deepfake Detection on Edge Devices via Adaptive Bidirectional Quantization

Deepfake detection has become a fundamental component of modern media forensics. Despite significant progress in detection accuracy, most existing methods remain computationally intensive and parameter-heavy, limiting their deployment on resource-constrained edge devices that require real-time, on-site inference. This limitation is particularly critical in an era where mobile devices are extensively used for media-centric applications, including online payments, virtual meetings, and social networking. Meanwhile, due to the unique requirement of capturing extremely subtle forgery artifacts for deepfake detection, state-of-the-art quantization techniques usually underperform for such a challenging task. These fine-grained cues are highly sensitive to model compression and can be easily degraded during quantization, leading to noticeable performance drops. This challenge highlights the need for quantization strategies specifically designed to preserve the discriminative features essential for reliable deepfake detection. To address this gap, we propose DefakeQ, the first quantization framework tailored for deepfake detectors, enabling real-time deployment on edge devices. Our approach introduces a novel adaptive bidirectional compression strategy that simultaneously leverages feature correlations and eliminates redundancy, achieving an effective balance between model compactness and detection performance. Extensive experiments across five benchmark datasets and eleven state-of-the-art backbone detectors demonstrate that DeFakeQ consistently surpasses existing quantization and model compression baselines. Furthermore, we deploy DefakeQ on mobile devices in real-world scenarios, demonstrating its capability for real-time deepfake detection and its practical applicability in edge environments.

Xiangyu Li, Yujing Sun, Yuhang Zheng +2

Digital Trust Centre, Nanyang Technological University · School of Information Science and Technology, ShanghaiTech University

Adaptive QuantizationDeepfake DetectionNeural Network QuantizationOn-Device Inference
AskPDF