TOCO framework compresses neural networks based on tolerance analysis.
problem Deploying large neural networks on edge devices with limited resources.
method TOCO uses tolerance analysis to perform fine-grained compression, allowing flexibility to hardware changes.
result Fine-grained compression of neural networks on edge devices.
Paper optimizes neural architectures for multiple device constraints.
problem NAS ignores device constraints like latency and energy.
method Developed MONAS and DPP-Net for multi-objective optimization.
result Found Pareto-optimal architectures for various devices.
POET enables large neural network training on tiny devices with reduced energy.
problem Training large neural networks on memory-limited edge devices.
method Jointly optimizes rematerialization and paging for memory reduction, formulating an MILP for energy-efficient training.
result POET trains ResNet-18 and BERT within Cortex-M memory constraints, outperforming current methods in energy efficiency.
This paper optimizes deep neural networks for resource-constrained devices.
problem Efficient deployment of deep neural networks on resource-constrained devices.
method Across-stack optimization of CNNs using weight pruning, channel pruning, quantization, and parallel execution.
result Comprehensive Pareto curves for trade-offs between accuracy, execution time, and memory space.
Train one network for efficient deployment across many devices.
problem Efficient inference across diverse devices with minimal resource constraints.
method Once-for-All (OFA) network training and progressive shrinking algorithm.
result OFA network achieves state-of-the-art accuracy with significantly reduced training time and resource usage.
The paper optimizes neural network inference on mobile GPUs.
problem Limited computing power and thermal constraints on mobile CPUs.
method Leverage mobile GPUs for neural network inference.
result Real-time inference of deep neural networks on Android and iOS devices.
New method reduces mobile device energy for image classification.
problem High energy consumption of CNNs on mobile devices.
method Global optimization of CNN architectures with Bayesian optimization.
result Reduces energy consumption by up to 6x compared to blackbox CNNs.
SEFR is a fast, energy-efficient classifier for ultra-low power devices.
problem Running machine learning on battery-powered devices is challenging due to time and energy constraints.
method SEFR is an ultra-low power classifier with linear time complexity for training and testing.
result SEFR is 63 times faster and 70 times more energy efficient than state-of-the-art classifiers.
Novel NAS method balances performance and hardware metrics efficiently.
problem Challenging multi-objective optimization in neural architecture search.
method Parameterizes joint architectural distribution via hypernetwork conditioned on hardware features and preferences.
result Zero-shot transferability to new devices with representative and diverse architectures.
Single-Path NAS designs efficient ConvNets for mobile devices in hours.
problem Designing efficient ConvNets for mobile devices under latency constraints.
method Single-Path NAS, a differentiable method that reduces search cost and inference time.
result Achieves state-of-the-art accuracy on ImageNet with 79ms inference latency.
FedDANE adapts DANE for federated learning, but underperforms compared to existing methods.
problem Federated learning's practical constraints and device heterogeneity.
method Adapted DANE for federated learning, providing convergence guarantees for convex and non-convex functions.
result Empirically, FedDANE underperforms compared to FedAvg and FedProx.
A novel approach for federated learning over-the-air computation to reduce latency and improve privacy.
problem Low-latency and privacy issues in edge machine learning for intelligent devices.
method Over-the-air computation and sparse-low-rank optimization for efficient global model aggregation.
result Efficient global model aggregation with low latency and improved privacy.
Plasma toroidal metric singularities in helical devices and tokamaks, giving rise to magnetic surfaces inside the plasma devices are investigated in two cases. In the first we consider the case of a rotational plasma on an helical device with circular cross-section and dissipation. In this case singularities are shown …
EC2T creates sparse and ternary neural networks for resource-constrained devices.
problem Deploying deep neural networks on resource-constrained devices.
method Entropy-Constrained Trained Ternarization (EC2T) framework.
result EC2T creates sparse and ternary neural networks that are efficient in terms of storage and computation.
Deep neural networks require large amounts of resources which makes them hard to use on resource constrained devices such as Internet-of-things devices. Offloading the computations to the cloud can circumvent these constraints but introduces a privacy risk since the operator of the cloud is not necessarily trustworthy.…
Edge language models show bias over time, especially on resource-constrained devices.
problem Bias in edge language models on resource-constrained devices.
method Comparative analysis of text-based bias across edge, cloud, and desktop environments; optimized Llama-2 model on Raspberry Pi 4; feedback loop mechanism to correct bias.
result Llama-2 on Raspberry Pi 4 shows 43.23% and 21.89% more bias over time compared to cloud and desktop models.
LAD-BNet improves real-time energy forecasting on edge devices.
problem Real-time energy forecasting on edge devices for smart grid optimization and intelligent buildings.
method Hybrid neural architecture combining temporal lag exploitation and TCN with dilated convolutions.
result 14.49% MAPE at 1-hour horizon with 18ms inference time on Edge TPU, 8-12x faster than CPU.
A DRL approach optimizes resource allocation in BFL to reduce latency and energy consumption.
problem Energy and CPU constraints of mobile devices and increased training latency due to blockchain mining.
method Deep Reinforcement Learning (DRL) to derive optimal decisions for MLMO.
result Optimal resource allocation and block generation rate to minimize system latency, energy consumption, and incentive cost.
Review of efficient neural networks for TinyML on resource-constrained devices.
problem Resource constraints on ultra-low power MCUs for deep learning models.
method Model compression, quantization, low-rank factorization, model pruning, hardware acceleration, algorithm-architecture co-design.
result Optimized neural network architectures for minimal resource utilization on MCUs.
Telemonitoring of electroencephalogram (EEG) through wireless body-area networks is an evolving direction in personalized medicine. Among various constraints in designing such a system, three important constraints are energy consumption, data compression, and device cost. Conventional data compression methodologies, al…
SPINN optimizes neural network inference on devices and cloud.
problem Inference on mobile devices is challenging due to high computational demands and dynamic connectivity.
method Synergistic progressive inference with a novel scheduler.
result SPINN achieves up to 2x higher throughput and reduces server cost by up to 6.8x.
Edge devices adapt pre-trained models to local data without backpropagation.
problem Adapting pre-trained models to edge devices' local data distributions.
method Feed-forward latent domain adaptation using cross-attention.
result Consistent improvements over ERM baselines and domain-supervised adaptation.
This paper compresses RNNs for IoT devices by 15-38x using Kronecker products.
problem Resource constraints on IoT devices make RNNs difficult to deploy.
method Kronecker product (KP) for compressing RNN layers by 15-38x with minimal accuracy loss.
result Kronecker product can compress RNNs by 50x when quantized to 8-bits.
Paper proposes a multi-phase pruning pipeline for deep ensemble learning on IIoT devices.
problem Computational limitations of IoT devices for deep learning models.
method Generates diverse pruned models, applies integer quantization, and uses clustering-based pruning.
result Significant reduction in model size (up to 90%) and improved performance (up to 7%) on IIoT devices.
This paper enables deep network inference on microcontrollers with improved accuracy and reduced memory usage.
problem Deploying deep networks on resource-constrained edge-devices with low memory and computational constraints.
method Mixed low-bitwidth compression, rule-based iterative procedure for bit precision determination, quantization-aware retraining, and integer-only model conversion.
result Improved Top1 accuracy of 68% on a 2MB FLASH memory STM32H7 microcontroller, 8% higher than 8-bit implementations.
Two new methods reduce communication costs in federated learning.
problem Heavy communication costs in federated learning, especially with non-IID data.
method FedMMD uses MMD constraint for two-stream model training; FedFusion aggregates local and global features.
result FedMMD and FedFusion reduce communication costs by 20% and 60% respectively.
Convex regression is a promising area for bridging statistical estimation and deterministic convex optimization. New piecewise linear convex regression methods are fast and scalable, but can have instability when used to approximate constraints or objective functions for optimization. Ensemble methods, like bagging, sm…
CoCoPIE shows AI can run on regular devices without special hardware.
problem Real-time AI on mobile devices without special hardware.
method Compression-compiler co-design using CoCoPIE framework.
result Real-time AI applications on off-the-shelf mobile devices.
New eGRU unit improves keyword spotting on ultra-low-power devices.
problem Resource constraints of edge devices for neural network deployment.
method Optimized recurrent unit architecture for ultra-low power.
result eGRU is 60x faster and 10x smaller than GRU, maintaining accuracy.
A new framework reduces data upload for image classification while protecting user privacy.
problem Data upload limitations and privacy concerns in cloud-based image classification.
method Unsupervised autoencoder training at edge devices, followed by latent vector transmission to server for classifier training.
result The framework reduces communications overhead and protects user data privacy.
MDLdroid improves mobile deep learning for personal sensing with faster training.
problem Continuous local changes and resource constraints in personal mobile sensing affect global model performance.
method ChainSGD-reduce approach to reduce overhead and balance resources.
result 2x to 3.5x faster training on off-the-shelf mobile devices compared to single-device training.
WEST compresses word embeddings and softmax layers for memory efficiency.
problem Memory constraints in large vocabulary models.
method WEST encodes words with sequences of sub-units, improving compression without performance loss.
result WEST achieves significant compression without sacrificing performance.
Optimizes pipelined computation and communication for edge learning within latency constraints.
problem Balancing data transmission and model training to meet latency requirements.
method Analyzes the optimal packet payload size tradeoff between bias and variance.
result Derives analytical bounds on the expected optimality gap for effective optimization.
Quantized neural networks reduce model size and energy consumption.
problem Memory and energy constraints in mobile devices.
method Using integer or binary representations to store weights instead of 32-bit floats.
result Quantization can reduce model size and energy consumption without significantly compromising performance.
Paper optimizes FL communication efficiency with stochastic optimization.
problem Intermittent connectivity and non-i.i.d. data in FL.
method Convergence analysis of non-convex loss functions, stochastic optimization for client selection and power allocation.
result Significant reduction in communication time compared to random participation.
Improved differentially private deep learning with group-wise clipping techniques.
problem Efficiency and privacy trade-offs in deep learning models.
method Group-wise clipping techniques (per-layer and per-device) to reduce compute time and memory overhead.
result Private learning with group-wise clipping achieves similar or better performance than non-private learning with less wall time.
Paper presents privacy-preserving techniques for HD computing.
problem Privacy loss in HD computing due to reversible computation.
method Quantization and pruning of hypervectors for differential privacy.
result Differentially private HD model for cloud inference.
New algorithm reduces communication time in federated learning.
problem Intermittent connectivity and non-i.i.d. data slow federated learning convergence.
method Lyapunov optimization for efficient device scheduling.
result Significant reduction in communication time with improved convergence rates.
Serenity optimizes neural network execution for edge devices by scheduling with optimal memory footprint.
problem Order of nodes in irregular neural networks affects memory footprint, complicating execution under resource constraints.
method Memory-aware compiler using dynamic programming and graph rewriting to find optimal schedules.
result Achieves optimal peak memory and further improves it with graph rewriting, reducing memory usage by 1.68x-1.86x compared to TensorFlow Lite.
New method prunes neural networks while conserving resources.
problem Design efficient neural networks for edge devices with limited computational power.
method Iterative channel-wise pruning with holonomic constraint in Lagrangian multipliers framework.
result Reduction of 15.47 GMAC to 3.87 GMAC for VGG-16 with only 1% top-1 accuracy loss.
TASO optimizes CNN models for memory-constrained devices.
problem Large CNN models are too resource-intensive for mobile devices.
method Integer Linear Programming (ILP) for selecting operations and workspace allocation.
result 8x speedup and 2.2x reduction in memory usage compared to greedy algorithms.
FedProx tackles heterogeneity in federated learning networks.
problem Significant variability in systems characteristics and non-identically distributed data in federated networks.
method FedProx is a framework that generalizes and re-parametrizes FedAvg, introducing modifications to handle both systems and statistical heterogeneity.
result FedProx demonstrates significantly more stable and accurate convergence behavior than FedAvg, improving test accuracy by 22% on average in highly heterogeneous settings.
Automates design of lightweight neural networks for image classification.
problem Designing efficient neural networks for edge devices with limited computational resources.
method Uses the Mesh Adaptive Direct Search (MADS) algorithm to optimize network architecture.
result Achieves comparable performance to standard methods with fewer design trials.
Paper proposes efficient pruning method for neural networks.
problem Compressing deep neural networks for resource-constrained devices.
method Adaptive sparsity loss for budget-aware optimization during training.
result Demonstrated effectiveness on various architectures and datasets.
Deep learning improves demosaicing but edge devices struggle.
problem Edge devices struggle with deep learning-based demosaicing.
method Exhaustive search of deep neural network architectures to find the best balance between performance and model complexity.
result Found architectures that outperform state-of-the-art demosaicing models on edge devices.
DFTerNet improves human activity recognition on portable devices with 2-bit quantization and dynamic fusion.
problem Resource constraints and unequal fusion strategies limit practical human activity recognition on portable devices.
method 2-bit Convolutional Neural Networks with dynamic fusion strategies for different activity types.
result Exceeds baseline model performance by up to ~5% on OPPORTUNITY and PAMAP2 datasets.
FedZKT enables resource-constrained devices to participate in federated learning with heterogeneous models.
problem Inequality in resource allocation hinders participation from resource-constrained devices in federated learning.
method Zero-shot knowledge transfer through a server-assigned distillation process.
result FedZKT effectively transfers knowledge across heterogeneous on-device models without requiring comparable local training efforts.
Artemis framework improves distributed learning with bidirectional compression and partial participation.
problem Learning in distributed or federated settings with communication constraints and device partial participation.
method Artemis framework using bidirectional compression, memory mechanism, and Polyak-Ruppert averaging.
result Fast rates of convergence (linear up to a threshold) under weak assumptions on stochastic gradients.