Limited Computational Resources Research Articles

DNN acceleration in vehicle edge computing with mobility-awareness: A synergistic vehicle–edge and edge–edge framework

In recent years, vehicular networks have seen a proliferation of applications and services such as image tagging, lane detection, and speech recognition. Many of these applications rely on Deep Neural Networks (DNNs) and demand low-latency computation. To meet these requirements, Vehicular Edge Computing (VEC) has been introduced to augment the abundant computation capacity of vehicular networks to complement limited computation resources on vehicles. Nevertheless, offloading DNN tasks to MEC (Multi-access Edge Computing) servers effectively and efficiently remains a challenging topic due to the dynamic nature of vehicular mobility and varying loads on the servers. In this paper, we propose a novel and efficient distributed DNN Partitioning And Offloading (DPAO), leveraging the mobility of vehicles and the synergy between vehicle–edge and edge–edge computing. We exploit the variations in both computation time and output data size across different layers of DNN to make optimized decisions for accelerating DNN computations while reducing the transmission time of intermediate data. In the meantime, we dynamically partition and offload tasks between MEC servers based on their load differences. We have conducted extensive simulations and testbed experiments to demonstrate the effectiveness of DPAO. The evaluation results show that, compared to offloaded all tasks to MEC server, DPAO reduces the latency of DNN tasks by 2.4x. DPAO with queue reservation can further reduce the task average completion time by 10%.

ARLP: Automatic multi-agent transformer reinforcement learning pruner for one-shot neural network pruning

Overparameterized Neural Networks demonstrate state-of-the-art performance; however, the escalating demand for more compact and energy-efficient neural networks has arisen to facilitate the deployment of machine learning applications on devices with limited computational resources. A prevalent approach employs various pruning techniques. However, hyperparameters, such as the pruning ratio for each layer in pruning techniques, are typically set by human experts and often lack optimization. In this paper, we therefore propose a novel method named “Automatic Multi-Agent Transformer Reinforcement Learning Pruner” (ARLP). ARLP leverages a transformer-based multi-agent reinforcement learning controller to autonomously prune networks at initialization by extracting network meta-features. This autonomous process eliminates the need for human intervention in determining the optimal pruning ratio for each layer. The meta-features are derived from various zero-cost pruning-at-initialization proxies to perform One-shot Pruning. Extensive experimental results demonstrate that ARLP outperforms other state-of-the-art methods, establishing its efficacy in achieving superior performance.

Democratizing protein language models with parameter-efficient fine-tuning

Proteomics has been revolutionized by large protein language models (PLMs), which learn unsupervised representations from large corpora of sequences. These models are typically fine-tuned in a supervised setting to adapt the model to specific downstream tasks. However, the computational and memory footprint of fine-tuning (FT) large PLMs presents a barrier for many research groups with limited computational resources. Natural language processing has seen a similar explosion in the size of models, where these challenges have been addressed by methods for parameter-efficient fine-tuning (PEFT). In this work, we introduce this paradigm to proteomics through leveraging the parameter-efficient method LoRA and training new models for two important tasks: predicting protein-protein interactions (PPIs) and predicting the symmetry of homooligomer quaternary structures. We show that these approaches are competitive with traditional FT while requiring reduced memory and substantially fewer parameters. We additionally show that for the PPI prediction task, training only the classification head also remains competitive with full FT, using five orders of magnitude fewer parameters, and that each of these methods outperform state-of-the-art PPI prediction methods with substantially reduced compute. We further perform a comprehensive evaluation of the hyperparameter space, demonstrate that PEFT of PLMs is robust to variations in these hyperparameters, and elucidate where best practices for PEFT in proteomics differ from those in natural language processing. All our model adaptation and evaluation code is available open-source at https://github.com/microsoft/peft_proteomics. Thus, we provide a blueprint to democratize the power of PLM adaptation to groups with limited computational resources.

Prior-free 3D tracking of a fast-moving object at 6667 frames per second with single-pixel detectors.

Real-time tracking and 3D trajectory computation of fast-moving objects is a promising technology, especially in the field of autonomous driving. However, existing image-based tracking methods face significant challenges when it comes to real-time tracking, primarily due to the limitation of storage space and computational resources. Here, we propose a novel approach that enables real-time 3D tracking of a fast-moving object without any prior motion information and at a very low computational cost. To enable 3D coordinate synthesis with a space-efficient optical setup, geometric moment patterns are projected on two non-orthogonal planes with a spatial resolution of 125 μm. Our experiment demonstrates an impressive tracking speed of 6667 frames per second (FPS) with a 20 kHz digital micromirror device (DMD), which is more than 200 times faster than the widely adopted video-based tracking methods. To the best of our knowledge, this is the highest tracking speed record in the field of single-pixel 3D trajectory tracking. This method promotes the development of real-time tracking techniques with single-pixel imaging (SPI).

Dynamic resource management for 6G vehicular networks: CORA-6G offloading and allocation strategies

Given the burgeoning demands of real-time applications and limited on-board computational resources, the rapid proliferation of 6G-enabled vehicular networks brings forth the critical challenge of efficient computation offloading. This paper introduces the CORA-6G algorithm, a novel solution specifically designed to address this issue by determining the optimal offloading ratio. Using a robust convex optimization problem, CORA-6G maximizes the computational utility, ensuring efficient resource allocation and enhanced responsiveness. Quantitative evaluations reveal that, when compared to conventional methods, CORA-6G improves offloading efficiency by up to 25%, thereby significantly boosting the overall performance of vehicular applications. This research underscores the potential of CORA-6G as a cornerstone for future 6G vehicular network advancements.

DESIGN: Online Device Selection and Edge Association for Federated Synergy Learning-enabled AIoT

The Artificial Intelligence of Things (AIoT) is an emerging technology that enables numerous AIoT devices to participate in big data analytics and machine learning (ML) model training, providing various customized intelligent services for industry manufacturing. Federated Learning (FL) empowers AIoT applications with privacy-preserving distributed model training without sharing raw data. However, due to IoT devices’ limited computing and memory resources, existing FL approaches for AIoT applications cannot support efficient large-scale model training. Federated synergy learning (FSyL) is a promising collaborative paradigm that alleviates the computation and communication overhead on resource-constrained AIoT devices via offloading part of the ML model to the edge server for end-to-edge collaborative training. Existing FSyL works neither efficiently address the inter-round device selection to improve model diversity nor determine the intra-round edge association to reduce the training cost, which hinders the applications of FSyL-enable AIoT. Motivated by this issue, this paper first investigates the bottlenecks of executing FSyL in AIoT. It builds an optimization model of joint inter-round device selection and intra-round edge association for balancing model diversity and training cost. To tackle the intractable coupling problem, we present a framework named Online DE vice S elect I on and Ed G e Associatio N for Cost-Diversity Trade-offs FSyL (DESIGN). First, the edge association subproblem is extracted from the original problem, and game theory determines the optimal association decision for an arbitrary device selection. Then, based on the optimal association decision, device selection is modeled as a combinatorial multi-armed bandit (CMAB) problem. Finally, we propose an online mechanism to obtain joint device selection and edge association decisions. The performance of DESIGN is theoretically analyzed and experimentally evaluated on real-world datasets. The results show that DESIGN can achieve up to \(84.3\%\) in cost-saving with an accuracy improvement of \(23.6\%\) compared with the state-of-the-art.

An improved problem transformation algorithm for large-scale multi-objective optimization

Solving Large-Scale Multi-Objective Optimization Problems (LSMOPs) is the major challenge in evolution computation. Due to the large number of decision variables involved, it is not easy for existing multi-objective evolutionary algorithms to search the entire decision variable space with limited computation resources. To address the above problem, a large-scale multi-objective evolutionary algorithm based on problem transformation, called LSMOEA/PT, is presented in this paper. LSMOEA/PT uses an improved problem transformation strategy to transform LSMOPs into small-scale multi-objective problems to accelerate convergence speed and reduce computational resources. Specifically, The transformation strategy selects reference solutions to initialize a weight population by applying the opposite individual method. After optimizing the weight population, a dual line strategy is implemented to combine the weight vectors with the decision variables from the original population, thereby generating a new population. The new population is combined with the original population for environment selection. In LSMOEA/PT, a double learning swarm optimizer is introduced to speed up the convergence of the population. A uniform distribution strategy is conducted to update transformed solutions after the problem transformation process, increasing the diversity of the population. Experimental results show that LSMOEA/PT is competitive with eleven state-of-the-art algorithms on large-scale multi-objective optimization test problems with up to 2000 decision variables.

Laplace-domain crosstalk-free source-encoded elastic full-waveform inversion using time-domain solvers

Crosstalk-free source-encoded elastic full-waveform inversion (FWI) using time-domain solvers demonstrates skill and efficiency at conducting seismic inversions involving multiple sources and receivers with limited computational resources. A drawback of common formulations of the procedure is that, by sweeping through the frequency domain randomly at a rate of one or a few sparsely sampled frequencies per shot, it is difficult to simultaneously incorporate time-selective data windows, as necessary for the targeting of arrivals or wave packets during the various stages of the inversion. Here, we solve this problem by using the Laplace transform of the data. Using complex-valued frequencies allows for damping the records with flexible decay rates and temporal offsets that target specific traveltimes. We present the theory of crosstalk-free source-encoded FWI in the Laplace domain, develop the details of its implementation, and illustrate the procedure with numerical examples relevant to exploration-scale scenarios.

EUAVDet: An Efficient and Lightweight Object Detector for UAV Aerial Images with an Edge-Based Computing Platform

Crafting an edge-based real-time object detector for unmanned aerial vehicle (UAV) aerial images is challenging because of the limited computational resources and the small size of detected objects. Existing lightweight object detectors often prioritize speed over detecting extremely small targets. To better balance this trade-off, this paper proposes an efficient and low-complexity object detector for edge computing platforms deployed on UAVs, termed EUAVDet (Edge-based UAV Object Detector). Specifically, an efficient feature downsampling module and a novel multi-kernel aggregation block are first introduced into the backbone network to retain more feature details and capture richer spatial information. Subsequently, an improved feature pyramid network with a faster ghost module is incorporated into the neck network to fuse multi-scale features with fewer parameters. Experimental evaluations on the VisDrone, SeaDronesSeeV2, and UAVDT datasets demonstrate the effectiveness and plug-and-play capability of our proposed modules. Compared with the state-of-the-art YOLOv8 detector, the proposed EUAVDet achieves better performance in nearly all the metrics, including parameters, FLOPs, mAP, and FPS. The smallest version of EUAVDet (EUAVDet-n) contains only 1.34 M parameters and achieves over 20 fps on the Jetson Nano. Our algorithm strikes a better balance between detection accuracy and inference speed, making it suitable for edge-based UAV applications.

An adaptive simulated annealing-based computational offloading scheme in UAV-assisted MEC networks

Mobile edge computing (MEC) is an advanced technology that enables fifth-generation (5G) applications. It uses a mechanism to offload computation from mobile devices (MDs) to edge servers at the access point (AP) to improve the quality of computation. By leveraging the strengths of unmanned aerial vehicles (UAVs), we can install edge servers on UAVs to support task offloading. However, this is constrained by the limited computational resources of UAVs and high energy consumption. Due to the complexity of such architecture, the joint optimization of energy consumption and delay by applying classical optimization algorithms cannot work well either. So far, there is no research work that addresses the problem of computation offloading in UAV-based MEC networks considering the UAV-AP-MD architecture that is well applicable in 5G scenarios; this motivates us to propose this work. We first formulate this problem as an optimization problem and then develop an adaptive simulated annealing-based method for joint optimization of energy consumption and delay. We design a mechanism to dynamically increase the cooling rate as a function of decreasing temperature to achieve better convergence, and implement a task queue for MDs and servers. Finally, we perform numerical simulations and find that the proposed method reduces the energy consumption (17%) and delay (50%) of MDs and UAV while it has the best task dropped rate (50%) and fairness (4.2%) compared to other known methods.

ESC-NAS: Environment Sound Classification Using Hardware-Aware Neural Architecture Search for the Edge

The combination of deep-learning and IoT plays a significant role in modern smart solutions, providing the capability of handling task-specific real-time offline operations with improved accuracy and minimised resource consumption. This study provides a novel hardware-aware neural architecture search approach called ESC-NAS, to design and develop deep convolutional neural network architectures specifically tailored for handling raw audio inputs in environmental sound classification applications under limited computational resources. The ESC-NAS process consists of a novel cell-based neural architecture search space built with 2D convolution, batch normalization, and max pooling layers, and capable of extracting features from raw audio. A black-box Bayesian optimization search strategy explores the search space and the resulting model architectures are evaluated through hardware simulation. The models obtained from the ESC-NAS process achieved the optimal trade-off between model performance and resource consumption compared to the existing literature. The ESC-NAS models achieved accuracies of 85.78%, 81.25%, 96.25%, and 81.0% for the FSC22, UrbanSound8K, ESC-10, and ESC-50 datasets, respectively, with optimal model sizes and parameter counts for edge deployment.

An adaptive strategy based multi-population multi-objective optimization algorithm

An algorithm is sensitive to parameters; different parameter settings for solving optimization problems can thus have a serious impact on algorithm performance. This leads to an inability to determine the optimal set of parameters for the algorithm to solve the problem at hand. In this study, we propose an adaptive strategy with a multi-population multi-objective algorithm (A-MPMO) framework to select the appropriate set of genetic settings according to the problem to be solved and eliminate the sensitivity of the algorithm to the parameters. Multi-population are often combined with Evolutionary Algorithms (EAs) as an effective strategy to maintain population diversity. First, we divided the population generated by the algorithm into multiple subpopulations to expand the search range and updated them iteratively using operators with different genetic parameters. Second, based on multi-population, subpopulations compete for limited computational resources, implying that the size of each subpopulation adaptively adjusts according to the degree of its contribution to problem solving. Finally, a set of subpopulations that are best suited to solve the problem is selected to improve the adaptability to different problems. For DTLZ, ZDT, and UF, compared to the other algorithms, A-MPMO was experimentally shown to produce better performance.

A Novel Attention‐Based Layer Pruning Approach for Low‐Complexity Convolutional Neural Networks

Deep learning (DL) has been very successful for classifying images, detecting targets, and segmenting regions in high‐resolution images such as whole slide histopathology images. However, analysis of such high‐resolution images requires very high DL complexity. Several AI optimization techniques have been recently proposed that aim at reducing the complexity of deep neural networks and hence expedite their execution and eventually allow the use of low‐power, low‐cost computing devices with limited computation and memory resources. These methods include parameter pruning and sharing, quantization, knowledge distillation, low‐rank approximation, and resource efficient architectures. Rather than pruning network structures including filters, layers, and blocks of layers based on a manual selection of a significance metric such as l1‐norm and l2‐norm of the filter kernels, novel highly efficient AI‐driven DL optimization algorithms using variations of the squeeze and excitation in order to prune filters and layers of deep models such as VGG‐16 as well as eliminate filters and blocks of residual networks such as ResNet‐56 are introduced. The proposed techniques achieve significantly higher reduction in the number of learning parameters, the number of floating point operations, and memory space as compared to the‐state‐of‐the‐art methods.

Implementation of Lightweight Machine Learning-Based Intrusion Detection System on IoT Devices of Smart Homes

Smart home devices, also known as IoT devices, provide significant convenience; however, they also present opportunities for attackers to jeopardize homeowners’ security and privacy. Securing these IoT devices is a formidable challenge because of their limited computational resources. Machine learning-based intrusion detection systems (IDSs) have been implemented on the edge and the cloud; however, IDSs have not been embedded in IoT devices. To address this, we propose a novel machine learning-based two-layered IDS for smart home IoT devices, enhancing accuracy and computational efficiency. The first layer of the proposed IDS is deployed on a microcontroller-based smart thermostat, which uploads the data to a website hosted on a cloud server. The second layer of the IDS is deployed on the cloud side for classification of attacks. The proposed IDS can detect the threats with an accuracy of 99.50% at cloud level (multiclassification). For real-time testing, we implemented the Raspberry Pi 4-based adversary to generate a dataset for man-in-the-middle (MITM) and denial of service (DoS) attacks on smart thermostats. The results show that the XGBoost-based IDS detects MITM and DoS attacks in 3.51 ms on a smart thermostat with an accuracy of 97.59%.

SpanEffiDet: Span-Scale and Span-Path Feature Fusion for Object Detection

Lower versions of EfficientDet (such as D0, D1) have smaller network structures and parameter sizes, but lower detection accuracy. Higher versions exhibit higher accuracy, but the increase in network complexity poses challenges for real-time processing and hardware requirements. To meet the higher accuracy requirements under limited computational resources, this paper introduces SpanEffiDet based on the channel adaptive frequency filter (CAFF) and the Span-Path Bidirectional Feature Pyramid structure. Firstly, the CAFF module proposed in this paper realizes the frequency domain transformation of channel information through Fourier transform and effectively extracts the key features through semantic adaptive frequency filtering, thus, eliminating channel redundant information of EfficientNet. Simultaneously, the module has the ability to compute the weights across the channels and at fine granularity, and capture the detailed information of element features. Secondly, a two-way characteristic pyramid network multi-level cross-BIFPN, which can achieve multi-layer and multi-nodes, is proposed to build cross-level information transmission to incorporate both semantic and positional information of the target. This design enables the network to more effectively detect objects with significant size differences in complex environments. Finally, by introducing generalized focal Loss V2, reliable localization quality estimation scores are predicted from the distribution statistics of bounding boxes, thereby improving localization accuracy. The experimental results indicate that on the MS COCO dataset, SpanEffiDet-D0 achieved an AP improvement of 3.3% compared to the original EfficientDet series algorithms. Similarly, on the PASCAL VOC2007 and 2012 datasets, the mAP of SpanEffiDet-D0 is respectively 1.66 and 2.65% higher than that of EfficientDet-D0.

Deep learning and quantization for accurate and efficient multi-target radar inference of moving targets

Real-time, radar-based human activity and target classification is useful for wide-area ground surveillance. However, the feasibility of deploying deep learning (DL) models in radar-based systems with limited computational resources remains unexplored. This paper investigated the effect of quantization on model throughput and accuracy for deployment in radar systems. A seven-layer residual network was proposed to classify ground-moving targets and achieved a test accuracy of 87.72%. The model was then quantized to 16-bit and 8-bit precision, resulting in a 3.8 times speedup in inference throughput, with less than a 0.4% drop in test and validation accuracy. The results showed that quantization can improve inference throughput with a negligible decrease in target classification accuracy. The increase in throughput and reduction in computational expense that comes with quantization promotes the feasibility of the deployment of DL models in systems with limited computational resources. The findings of this paper hold significant promise for the successful use of quantized models in modern radar systems, while adhering to stringent size, weight and power consumption constraints.

Hybridizing traditional and next-generation reservoir computing to accurately and efficiently forecast dynamical systems.

Reservoir computers (RCs) are powerful machine learning architectures for time series prediction. Recently, next generation reservoir computers (NGRCs) have been introduced, offering distinct advantages over RCs, such as reduced computational expense and lower training data requirements. However, NGRCs have their own practical difficulties, including sensitivity to sampling time and type of nonlinearities in the data. Here, we introduce a hybrid RC-NGRC approach for time series forecasting of dynamical systems. We show that our hybrid approach can produce accurate short-term predictions and capture the long-term statistics of chaotic dynamical systems in situations where the RC and NGRC components alone are insufficient, e.g., due to constraints from limited computational resources, sub-optimal hyperparameters, sparsely sampled training data, etc. Under these conditions, we show for multiple model chaotic systems that the hybrid RC-NGRC method with a small reservoir can achieve prediction performance approaching that of a traditional RC with a much larger reservoir, illustrating that the hybrid approach can offer significant gains in computational efficiency over traditional RCs while simultaneously addressing some of the limitations of NGRCs. Our results suggest that the hybrid RC-NGRC approach may be particularly beneficial in cases when computational efficiency is a high priority and an NGRC alone is not adequate.

Enhancing wireless sensor network security with optimized cluster head selection and hybrid public-key encryption

This paper introduces an integrated methodology that enhances both the efficiency and security of wireless sensor networks (WSNs) against various active attacks. A two-fold strategy is proposed that incorporates an advanced cluster head (CH) selection and a customized, lightweight encryption protocol. The CH selection process is optimized through a multi-phase approach using fuzzy logic, local and global network qualifiers, and a trust index to ensure the election of CHs that are not only energy-efficient but also reliable. To complement the robust CH selection, the study introduces a hybrid yet lightweight encryption scheme customized Rivest-Shamir-Adleman (c-RSA) and customized advanced encryption standard (c-AES) algorithms. This scheme is customized for WSNs with limited computational resources, maintaining strong encryption standards while significantly reducing energy consumption and computational overhead. Experimental results demonstrate that the proposed system substantially enhances network performance, exhibiting a 34.15% improvement in energy efficiency and a 30.95% increase in reliability over existing methods such as LEACH and its modified versions. This comprehensive approach underscores the potential for a synergistic design in WSNs that does not compromise on security while optimizing operational efficiency.

Quantized non-volatile nanomagnetic domain wall synapse based autoencoder for efficient unsupervised network anomaly detection

Anomaly detection in real-time using autoencoders implemented on edge devices is exceedingly challenging due to limited hardware, energy, and computational resources. We show that these limitations can be addressed by designing an autoencoder with low-resolution non-volatile memory-based synapses and employing an effective quantized neural network learning algorithm. We further propose nanoscale ferromagnetic racetracks with engineered notches hosting magnetic domain walls (DW) as exemplary non-volatile memory-based autoencoder synapses, where limited state (5-state) synaptic weights are manipulated by spin orbit torque (SOT) current pulses to write different magnetoresistance states. The performance of anomaly detection of the proposed autoencoder model is evaluated on the NSL-KDD dataset. Limited resolution and DW device stochasticity aware training of the autoencoder is performed, which yields comparable anomaly detection performance to the autoencoder having floating-point precision weights. While the limited number of quantized states and the inherent stochastic nature of DW synaptic weights in nanoscale devices are typically known to negatively impact the performance, our hardware-aware training algorithm is shown to leverage these imperfect device characteristics to generate an improvement in anomaly detection accuracy (90.98%) compared to accuracy obtained with floating-point synaptic weights that are extremely memory intensive. Furthermore, our DW-based approach demonstrates a remarkable reduction of at least three orders of magnitude in weight updates during training compared to the floating-point approach, implying significant reduction in operation energy for our method. This work could stimulate the development of extremely energy efficient non-volatile multi-state synapse-based processors that can perform real-time training and inference on the edge with unsupervised data.

Multitask Learning for Joint Diagnosis of Multiple Mental Disorders in Resting-State fMRI.

Facing the increasing worldwide prevalence of mental disorders, the symptom-based diagnostic criteria struggle to address the urgent public health concern due to the global shortfall in well-qualified professionals. Thanks to the recent advances in neuroimaging techniques, functional magnetic resonance imaging (fMRI) has surfaced as a new solution to characterize neuropathological biomarkers for detecting functional connectivity (FC) anomalies in mental disorders. However, the existing computer-aided diagnosis models for fMRI analysis suffer from unstable performance on large datasets. To address this issue, we propose an efficient multitask learning (MTL) framework for joint diagnosis of multiple mental disorders using resting-state fMRI data. A novel multiobjective evolutionary clustering algorithm is presented to group regions of interests (ROIs) into different clusters for FC pattern analysis. On the optimal clustering solution, the multicluster multigate mixture-of-expert model is used for the final classification by capturing the highly consistent feature patterns among related diagnostic tasks. Extensive simulation experiments demonstrate that the performance of the proposed framework is superior to that of the other state-of-the-art methods. Moreover, the potential for practical application of the framework is also validated in terms of limited computational resources, real-time analysis, and insufficient training data. The proposed model can identify the remarkable interpretative biomarkers associated with specific mental disorders for clinical interpretation analysis.