Search

Due to the increasing size of integrated circuits (ICs), their design and optimization phases (i.e., computer-aided design, CAD) grow increasingly complex. At design time, a large design space needs to be explored to find an implementation that fulfills all specifications and then optimizes metrics like energy, area, delay, reliability, etc. At run time, a large configuration space needs to be searched to find the best set of parameters (e.g., voltage/frequency) to further optimize the system. Both spaces are infeasible for exhaustive search typically leading to heuristic optimization algorithms that find some tradeoff between design quality and computational overhead. Machine learning (ML) can build powerful models that have successfully been employed in related domains. In this survey, we categorize how ML may be used and is used for design-time and run-time optimization and exploration strategies of ICs. A metastudy of published techniques unveils areas in CAD that are well explored and underexplored with ML, as well as trends in the employed ML algorithms. We present a comprehensive categorization and summary of the state of the art on ML for CAD. Finally, we summarize the remaining challenges and promising open research directions. [ABSTRACT FROM AUTHOR]

This paper investigates the stochastic optimization problem focusing on developing scalable parallel algorithms for deep learning tasks. Our solution involves a reformation of the objective function for stochastic optimization in neural network models, along with a novel parallel computing strategy, coined the weighted aggregating stochastic gradient descent (WASGD). Following a theoretical analysis on the characteristics of the new objective function, WASGD introduces a decentralized weighted aggregating scheme based on the performance of local workers. Without any center variable, the new method automatically gauges the importance of local workers and accepts them by their contributions. Furthermore, we have developed an enhanced version of the method, WASGD+, by (1) implementing a designed sample order and (2) upgrading the weight evaluation function. To validate the new method, we benchmark our pipeline against several popular algorithms including the state-of-the-art deep neural network classifier training techniques (e.g., elastic averaging SGD). Comprehensive validation studies have been conducted on four classic datasets: CIFAR-100, CIFAR-10, Fashion-MNIST, and MNIST. Subsequent results have firmly validated the superiority of the WASGD scheme in accelerating the training of deep architecture. Better still, the enhanced version, WASGD+, is shown to be a significant improvement over its prototype. [ABSTRACT FROM AUTHOR]

Deep convolutional neural networks (DCNNs) are currently the method of choice both for generative, as well as for discriminative learning in computer vision and machine learning. The success of DCNNs can be attributed to the careful selection of their building blocks (e.g., residual blocks, rectifiers, sophisticated normalization schemes, to mention but a few). In this paper, we propose $\Pi$ Π -Nets, a new class of function approximators based on polynomial expansions. $\Pi$ Π -Nets are polynomial neural networks, i.e., the output is a high-order polynomial of the input. The unknown parameters, which are naturally represented by high-order tensors, are estimated through a collective tensor factorization with factors sharing. We introduce three tensor decompositions that significantly reduce the number of parameters and show how they can be efficiently implemented by hierarchical neural networks. We empirically demonstrate that $\Pi$ Π -Nets are very expressive and they even produce good results without the use of non-linear activation functions in a large battery of tasks and signals, i.e., images, graphs, and audio. When used in conjunction with activation functions, $\Pi$ Π -Nets produce state-of-the-art results in three challenging tasks, i.e., image generation, face verification and 3D mesh representation learning. The source code is available at https://github.com/grigorisg9gr/polynomial_nets. [ABSTRACT FROM AUTHOR]

Machine Learning (ML) workloads are increasingly deployed at the edge. Enabling efficient inference execution while considering model and system heterogeneity remains challenging, especially for ML tasks built with a network of deep neural networks (DNNs). The challenge is to maximize the utilization of all available resources on the multiprocessor system on a chip (MPSoC) at the same time. This becomes even more complicated because the optimal mapping for the network of DNNs can vary with input batch sizes and scene complexity. In this paper, a holistic hierarchical scheduling framework is presented to optimize the execution time for a network of DNN models on an edge MPSoC at runtime, considering varying input characteristics. The framework consists of a local and a global scheduler. The local scheduler maps individual DNNs in the inference pipeline to the best-performing hardware unit while the global scheduler customizes an Integer Linear Programming (ILP) solution to instantiate DNN remapping. To minimize scheduler runtime overhead, an imitation learning (IL) based scheduler is used that approximates the ILP solutions. The proposed scheduling framework (CAMDNN) was implemented on a Qualcomm Robotic RB5 platform. CAMDNN resulted in lower execution time of up to 32% than heterogeneous earliest finish time, and by factors of 6.67X, 5.6X and 2.17X than the CPU-only, GPU-only and Central Queue schedulers. [ABSTRACT FROM AUTHOR]

Parallelism and data reuse are the most critical issues for the design of hardware acceleration in a deep learning processor. Besides, abundant on-chip memories and precise data management are intrinsic design requirements because most of deep learning algorithms are data-driven and memory-bound. In this paper, we propose a compiler-architecture co-design scheme targeting a reconfigurable and algorithm-oriented array processor, named ReAAP. Given specific deep neural networks, the proposed co-design scheme is effective to perform parallelism and data reuse optimization on compute-intensive layers for guiding reconfigurable computing in hardware. Especially, the systemic optimization is performed in our proposed domain-specific compiler to deal with the intrinsic tensions between parallelism and data locality, for the purpose of automatically mapping diverse layer-level workloads onto our proposed reconfigurable array architecture. In this architecture, abundant on-chip memories are software-controlled and its massive data access is precisely handled by compiler-generated instructions. In our experiments, the ReAAP is implemented on an embedded FPGA platform. Experimental results demonstrate that our proposed co-design scheme is effective to integrate software flexibility with hardware parallelism for accelerating diverse deep learning workloads. As a whole system, ReAAP achieves a consistently high utilization of hardware resource for accelerating all the diverse compute-intensive layers in ResNet, MobileNet, and BERT. [ABSTRACT FROM AUTHOR]

Classification of 3D objects – the selection of a category in which each object belongs – is of great interest in the field of machine learning. Numerous researchers use deep neural networks to address this problem, altering the network architecture and representation of the 3D shape used as an input. To investigate the effectiveness of their approaches, we conduct an extensive survey of existing methods and identify common ideas by which we categorize them into a taxonomy. Second, we evaluate 11 selected classification networks on two 3D object datasets, extending the evaluation to a larger dataset on which most of the selected approaches have not been tested yet. For this, we provide a framework for converting shapes from common 3D mesh formats into formats native to each network, and for training and evaluating different classification approaches on this data. Despite being partially unable to reach the accuracies reported in the original papers, we compare the relative performance of the approaches as well as their performance when changing datasets as the only variable to provide valuable insights into performance on different kinds of data. We make our code available to simplify running training experiments with multiple neural networks with different prerequisites. [ABSTRACT FROM AUTHOR]

Hyperdimensional computing (HDC) has emerged as an alternative lightweight learning solution to deep neural networks. A key characteristic of HDC is the great extent of parallelism that can facilitate hardware acceleration. However, previous hardware implementations of HDC seldom focus on GPU designs, which were also inefficient partly due to the complexity of accelerating HDC on GPUs. In this paper, we present OpenHD, a flexible and high-performance GPU-powered framework for automating the mapping of general HDC applications including classification and clustering to GPUs. OpenHD takes advantage of memory optimization strategies specialized for HDC, minimizing the access time to different memory subsystems, and removing redundant operations. We also propose a novel training method to enable data parallelism in HDC training. Our evaluation result shows that the proposed training rapidly achieves the target accuracy, reducing the required training epochs by 4×. With OpenHD, users can deploy GPU-accelerated HDC applications without domain expert knowledge. Compared to the state-of-the-art GPU-powered HDC implementation, our evaluation on NVIDIA Jetson TX2 shows that OpenHD is up to 10.5× and 314× faster for HDC-based classification and clustering, respectively. Compared with non-HDC classification and clustering on GPUs, OpenHD-based HDC is 11.7× and 53× faster at comparable accuracy. OpenHD is available at: https://github.com/UCSD-SEELab/openhd. [ABSTRACT FROM AUTHOR]

The efficiency of link adaptation in wireless communications relies greatly on the accuracy of channel knowledge and transmission mode selection. In this paper, a novel deep learning based link adaptation framework is proposed for the orthogonal frequency-division multiplexing (OFDM) systems with compressed-sensing-assisted index modulation, termed as OFDM-CSIM, communicating over millimeter-wave (mmWave) channels. To achieve link adaptation, a novel multi-layer sparse Bayesian learning (SBL) algorithm is proposed for accurately and instantaneously providing the required channel state information. Meanwhile, a deep neural networks (DNN)-assisted adaptive modulation algorithm is proposed to choose the best possible transmission mode to maximize the achievable throughput. Simulation results show that the proposed multi-layer SBL algorithm enables more accurate channel estimation than the conventional techniques. The DNN-based adaptive modulator is capable of achieving a higher throughput than the learning-assisted solution based on the $k$ nearest neighbor ($k$ -NN) algorithm, and also the classic average signal-to-noise ratio (SNR)-based solutions. Moreover, analysis shows that both the multi-layer SBL algorithm and the DNN-assisted adaptive modulator achieve better performance than their respective conventional counterparts while at a significantly lower computational complexity cost. [ABSTRACT FROM AUTHOR]

With the popularity of location-aware devices (e.g., smart phones), large amounts of location-based social media data such as check-ins are generated. This stimulates plenty of studies for POI recommendations by applying machine learning techniques. However, most of the existing studies focus on POI recommendations in the same city or region, and fail to recommend POIs for users when they travel to a new city. In this paper, we propose a novel deep neural network, named as ST-TransRec, for crossing-city POI recommendations. It integrates the deep neural network, transfer learning technique, and density-based resampling method into a unified framework. In this model, the deep neural network is used to capture users’ preferences for POIs and learn the embeddings of POIs. Besides, the transfer learning technique is employed to bridge the gap between cities that results from the city-dependent features. As the distributions over POIs are imbalanced, we design a density-based spatial resampling model which enables POIs to be well matched across cities. We conduct extensive experiments on two real-world datasets. The experimental results show the advantages of ST-TransRec over the state-of-the-art methods for crossing-city POI recommendations. [ABSTRACT FROM AUTHOR]

With the fast development of various positioning techniques such as Global Position System (GPS), mobile devices and remote sensing, spatio-temporal data has become increasingly available nowadays. Mining valuable knowledge from spatio-temporal data is critically important to many real-world applications including human mobility understanding, smart transportation, urban planning, public safety, health care and environmental management. As the number, volume and resolution of spatio-temporal data increase rapidly, traditional data mining methods, especially statistics-based methods for dealing with such data are becoming overwhelmed. Recently deep learning models such as recurrent neural network (RNN) and convolutional neural network (CNN) have achieved remarkable success in many domains due to the powerful ability in automatic feature representation learning, and are also widely applied in various spatio-temporal data mining (STDM) tasks such as predictive learning, anomaly detection and classification. In this paper, we provide a comprehensive review of recent progress in applying deep learning techniques for STDM. We first categorize the spatio-temporal data into five different types, and then briefly introduce the deep learning models that are widely used in STDM. Next, we classify existing literature based on the types of spatio-temporal data, the data mining tasks, and the deep learning models, followed by the applications of deep learning for STDM in different domains including transportation, on-demand service, climate & weather analysis, human mobility, location-based social network, crime analysis, and neuroscience. Finally, we conclude the limitations of current research and point out future research directions. [ABSTRACT FROM AUTHOR]

In this paper, we aim to efficiently and accurately predict the static channel impulse response (CIR) with only the user’s position information and a set of channel instances obtained within a certain wireless communication environment. Such a problem is by no means trivial since it needs to reconstruct the high-dimensional information (here the CIR everywhere) from the extremely low-dimensional data (here the location coordinates), which often results in overfitting and large prediction error. To this end, we resort to a novel physics-inspired generative approach. Specifically, we first use a forward deep neural network to infer the positions of all possible images of the source reflected by the surrounding scatterers within that environment, and then use the well-known Gaussian Radial Basis Function network (GRBF) to approximate the amplitudes of all possible propagation paths. We further incorporate the most recently developed sinusoidal representation network (SIREN) into the proposed network to implicitly represent the highly dynamic phases of all possible paths, which usually cannot be well predicted by the conventional neural networks with non-periodic activators. The resultant framework of Cosine-Gaussian Radial Basis Function network (C-GRBFnet) is also extended to the MIMO channel case. Key performance measures including prediction accuracy, convergence speed, network scale and robustness to channel estimation error are comprehensively evaluated and compared with existing popular networks, which show that our proposed network is much more efficient in representing, learning and predicting wireless channels in a given communication environment. [ABSTRACT FROM AUTHOR]

Monitoring the depth of unconsciousness during anesthesia is beneficial in both clinical settings and neuroscience investigations to understand brain mechanisms. Electroencephalogram (EEG) has been used as an objective means of characterizing brain altered arousal and/or cognition states induced by anesthetics in real-time. Different general anesthetics affect cerebral electrical activities in different ways. However, the performance of conventional machine learning models on EEG data is unsatisfactory due to the low Signal to Noise Ratio (SNR) in the EEG signals, especially in the office-based anesthesia EEG setting. Deep learning models have been used widely in the field of Brain Computer Interface (BCI) to perform classification and pattern recognition tasks due to their capability of good generalization and handling noises. Compared to other BCI applications, where deep learning has demonstrated encouraging results, the deep learning approach for classifying different brain consciousness states under anesthesia has been much less investigated. In this paper, we propose a new framework based on meta-learning using deep neural networks, named Anes-MetaNet, to classify brain states under anesthetics. The Anes-MetaNet is composed of Convolutional Neural Networks (CNN) to extract power spectrum features, and a time consequence model based on Long Short-Term Memory (LSTM) networks to capture the temporal dependencies, and a meta-learning framework to handle large cross-subject variability. We use a multi-stage training paradigm to improve the performance, which is justified by visualizing the high-level feature mapping. Experiments on the office-based anesthesia EEG dataset demonstrate the effectiveness of our proposed Anes-MetaNet by comparison of existing methods. [ABSTRACT FROM AUTHOR]

The segmentation of multiple sclerosis (MS) lesions from MR imaging sequences remains a challenging task, due to the characteristics of variant shapes, scattered distributions and unknown numbers of lesions. However, the current automated MS segmentation methods with deep learning models face the challenges of (1) capturing the scattered lesions in multiple regions and (2) delineating the global contour of variant lesions. To address these challenges, in this paper, we propose a novel attention and graph-driven network (DAG-Net), which incorporates (1) the spatial correlations for embracing the lesions in distant regions and (2) the global context for better representing lesions of variant features in a unified architecture. Firstly, the novel local attention coherence mechanism is designed to construct dynamic and expansible graphs for the spatial correlations between pixels and their proximities. Secondly, the proposed spatial-channel attention module enhances features to optimize the global contour delineation, by aggregating relevant features. Moreover, with the dynamic graphs, the learning process of the DAG-Net is interpretable, which in turns support the reliability of segmentation results. Extensive experiments were conducted on a public ISBI2015 dataset and an in-house dataset in comparison to state-of-the-art methods, based on geometrical and clinical metrics. The experimental results validate the effectiveness of proposed DAG-Net on segmenting variant and scatted lesions in multiple regions. [ABSTRACT FROM AUTHOR]

This paper presents a novel Transformer-based facial landmark localization network named Localization Transformer (LOTR). The proposed framework is a direct coordinate regression approach leveraging a Transformer network to better utilize the spatial information in a feature map. An LOTR model consists of three main modules: 1) a visual backbone that converts an input image into a feature map, 2) a Transformer module that improves the feature representation from the visual backbone, and 3) a landmark prediction head that directly predicts landmark coordinates from the Transformer’s representation. Given cropped-and-aligned face images, the proposed LOTR can be trained end-to-end without requiring any post-processing steps. This paper also introduces a loss function named smooth-Wing loss, which addresses the gradient discontinuity of the Wing loss, leading to better convergence than standard loss functions such as L1, L2, and Wing loss. Experimental results on the JD landmark dataset provided by the First Grand Challenge of 106-Point Facial Landmark Localization indicate the superiority of LOTR over the existing methods on the leaderboard and two recent heatmap-based approaches. On the WFLW dataset, the proposed LOTR framework demonstrates promising results compared with several state-of-the-art methods. Additionally, we report an improvement in the performance of state-of-the-art face recognition systems when using our proposed LOTRs for face alignment.

In this modern era, traffic congestion has become a major source of severe negative economic and environmental impact for urban areas worldwide. One of the most efficient ways to mitigate traffic congestion is through future traffic prediction. The research field of traffic prediction has evolved greatly ever since its inception in the late 70s. Earlier studies mainly use classical statistical models such as ARIMA and its variants. Recently, researchers have started to focus on machine learning models because of their power and flexibility. As theoretical and technological advances emerge, we enter the era of deep neural network, which gained popularity due to its sheer prediction power which can be attributed to the complex and deep structure. Despite the popularity of deep neural network models in the field of traffic prediction, literature surveys of such methods are rare. In this work, we present an up-to-date survey of deep neural network for traffic prediction. We will provide a detailed explanation of popular deep neural network architectures commonly used in the traffic flow prediction literatures, categorize and describe the literatures themselves, present an overview of the commonalities and differences among different works, and finally provide a discussion regarding the challenges and future directions for this field. [ABSTRACT FROM AUTHOR]

The emerging technology of reconfigurable intelligent surfaces (RISs) is provisioned as an enabler of smart wireless environments, offering a highly scalable, low-cost, hardware-efficient, and almost energy-neutral solution for dynamic control of the propagation of electromagnetic signals over the wireless medium, ultimately providing increased environmental intelligence for diverse operation objectives. One of the major challenges with the envisioned dense deployment of RISs in such reconfigurable radio environments is the efficient configuration of multiple metasurfaces with limited, or even the absence of, computing hardware. In this article, we consider multiuser and multi-RIS-empowered wireless systems and present a thorough survey of the online machine learning approaches for the orchestration of their various tunable components. Focusing on the sum-rate maximization as a representative design objective, we present a comprehensive problem formulation based on deep reinforcement learning (DRL). We detail the correspondences among the parameters of the wireless system and the DRL terminology, and devise generic algorithmic steps for the artificial neural network training and deployment while discussing their implementation details. Further practical considerations for multi-RIS-empowered wireless communications in the sixth-generation (6G) era are presented along with some key open research challenges. Different from the DRL-based status quo, we leverage the independence between the configuration of the system design parameters and the future states of the wireless environment, and present efficient multiarmed bandits approaches, whose resulting sum-rate performances are numerically shown to outperform random configurations, while being sufficiently close to the conventional deep $Q$ network (DQN) algorithm, but with lower implementation complexity. [ABSTRACT FROM AUTHOR]

With the aggressive and amazing scaling of the feature size of semiconductors, hotspot detection has become a crucial and challenging problem in the generation of optimized mask design for better printability. Machine learning techniques, especially deep learning, have attained notable success on hotspot detection tasks. However, most existing hotspot detectors suffer from suboptimal performance due to two-stage flow and less efficient representations of layout features. What is more, most works can only solve simple benchmarks with apparent hotspot patterns like ICCAD 2012 Contest benchmarks. In this article, we first develop a new end-to-end hotspot detection flow where layout feature embedding and hotspot detection are jointly performed. An attention mechanism-based deep convolutional neural network (CNN) is exploited as the backbone to learn embeddings for layout features and classify the hotspots simultaneously. The experimental results demonstrate that our framework achieves accuracy improvement over prior arts with fewer false alarms and faster inference speed on much more challenging benchmarks. [ABSTRACT FROM AUTHOR]

Deep neural networks (DNNs), regardless of their impressive performance, are vulnerable to attacks from adversarial inputs and, more recently, Trojans to misguide or hijack the decision of the model. We expose the existence of an intriguing class of spatially bounded, physically realizable, adversarial examples— Universal NaTuralistic adversarial paTches—we call TnTs, by exploring the super set of the spatially bounded adversarial example space and the natural input space within generative adversarial networks. Now, an adversary can arm themselves with a patch that is naturalistic, less malicious-looking, physically realizable, highly effective—achieving high attack success rates, and universal. A TnT is universal because any input image captured with a TnT in the scene will: i) misguide a network (untargeted attack); or ii) force the network to make a malicious decision (targeted attack). Interestingly, now, an adversarial patch attacker has the potential to exert a greater level of control—the ability to choose a location independent, natural-looking patch as a trigger in contrast to being constrained to noisy perturbations—an ability is thus far shown to be only possible with Trojan attack methods needing to interfere with the model building processes to embed a backdoor at the risk discovery; but, still realize a patch deployable in the physical world. Through extensive experiments on the large-scale visual classification task, ImageNet with evaluations across its entire validation set of 50,000 images, we demonstrate the realistic threat from TnTs and the robustness of the attack. We show a generalization of the attack to create patches achieving higher attack success rates than existing state-of-the-art methods. Our results show the generalizability of the attack to different visual classification tasks (CIFAR-10, GTSRB, PubFig) and multiple state-of-the-art deep neural networks such as WideResnet50, Inception-V3 and VGG-16. [ABSTRACT FROM AUTHOR]

Real-world datasets commonly have noisy labels, which negatively affects the performance of deep neural networks (DNNs). In order to address this problem, we propose a label noise robust learning algorithm, in which the base classifier is trained on soft-labels that are produced according to a meta-objective. In each iteration, before conventional training, the meta-training loop updates soft-labels so that resulting gradients updates on the base classifier would yield minimum loss on meta-data. Soft-labels are generated from extracted features of data instances, and the mapping function is learned by a single layer perceptron (SLP) network, which is called MetaLabelNet. Following, base classifier is trained by using these generated soft-labels. These iterations are repeated for each batch of training data. Our algorithm uses a small amount of clean data as meta-data, which can be obtained effortlessly for many cases. We perform extensive experiments on benchmark datasets with both synthetic and real-world noises. Results show that our approach outperforms existing baselines. The source code of the proposed model is available at https://github.com/gorkemalgan/MetaLabelNet. [ABSTRACT FROM AUTHOR]

Objective: We show that state-of-the-art deep neural networks achieve superior results in regression-based multi-class proportional myoelectric hand prosthesis control than two common baseline approaches, and we analyze the neural network mapping to explain why this is the case. Methods: Feedforward neural networks and baseline systems are trained on an offline corpus of 11 able-bodied subjects and 4 prosthesis wearers, using the $R^2$ score as metric. Analysis is performed using diverse qualitative and quantitative approaches, followed by a rigorous evaluation. Results: Our best neural networks have at least three hidden layers with at least 128 neurons per layer; smaller architectures, as used by many prior studies, perform substantially worse. The key to good performance is to both optimally regress the target movement, and to suppress spurious movements. Due to the properties of the underlying data, this is impossible to achieve with linear methods, but can be attained with high exactness using sufficiently large neural networks. Conclusion: Neural networks perform significantly better than common linear approaches in the given task, in particular when sufficiently large architectures are used. This can be explained by salient properties of the underlying data, and by theoretical and experimental analysis of the neural network mapping. Significance: To the best of our knowledge, this work is the first one in the field which not only reports that large and deep neural networks are superior to existing architectures, but also explains this result. [ABSTRACT FROM AUTHOR]

Despite its relatively recent history, deep learning (DL)-based source code analysis is already a cornerstone in machine learning for compiler optimization. When applied to the classification of pieces of code to identify the best computational unit in a heterogeneous Systems-on-Chip, it can be effective in supporting decisions that a programmer has otherwise to take manually. Several techniques have been proposed exploiting different networks and input information, prominently sequence-based and graph-based representations, complemented by auxiliary information typically related to payload and device configuration. While the accuracy of DL methods strongly depends on the training and test datasets, so far no exhaustive and statistically meaningful analysis has been done on its impact on the results and on how to effectively extract the available information. This is relevant also considering the scarce availability of source code datasets that can be labeled by profiling on heterogeneous compute units. In this article, we first present such a study, which leads us to devise the contribution of code sequences and auxiliary inputs separately. Starting from this analysis, we then demonstrate that by using the normalization of auxiliary information, it is possible to improve state-of-the-art results in terms of accuracy. Finally, we propose a novel approach exploiting Siamese networks that further improve mapping accuracy by increasing the cardinality of the dataset, thus compensating for its relatively small size. [ABSTRACT FROM AUTHOR]

Early fault detection in power electronic systems (PESs) to maintain reliability is one of the most important issues that has been significantly addressed in recent years. In this article, after reviewing various works of literature based on fault detection in PESs, data mining-based techniques including artificial neural network, machine learning, and deep learning algorithms are introduced. Then, the fault detection routine in PESs is expressed by introducing signal measurement sensors and how to extract the feature from them. Finally, based on studies, the performance of various data mining methods in detecting PESs faults is evaluated. The results of evaluations show that the deep learning-based techniques given the ability of feature extraction from measured signals are significantly more effective than other methods and as an ideal tool for future applications in the power electronics industry are introduced. [ABSTRACT FROM AUTHOR]

Federated learning is a distributed machine learning approach to privacy preservation and two major technical challenges prevent a wider application of federated learning. One is that federated learning raises high demands on communication resources, since a large number of model parameters must be transmitted between the server and clients. The other challenge is that training large machine learning models such as deep neural networks in federated learning requires a large amount of computational resources, which may be unrealistic for edge devices such as mobile phones. The problem becomes worse when deep neural architecture search (NAS) is to be carried out in federated learning. To address the above challenges, we propose an evolutionary approach to real-time federated NAS that not only optimizes the model performance but also reduces the local payload. During the search, a double-sampling technique is introduced, in which for each individual, only a randomly sampled submodel is transmitted to a number of randomly sampled clients for training. This way, we effectively reduce computational and communication costs required for evolutionary optimization, making the proposed framework well suitable for real-time federated NAS. [ABSTRACT FROM AUTHOR]

Recently, the use of keyword spotting (KWS) has become prevalent in mobile devices. State-of-the-art deep learning algorithms such as temporal convolutional networks (TCNs) have been applied to this task achieving superior accuracy results. These models can, however, be mapped in multiple ways onto embedded devices, ranging from real-time streaming inference with or without computational sprinting to delayed batched inference. Although functionally equivalent, these deployment settings, however, strongly impacts average power consumption and latency of this real time task, hence requiring a thorough optimization. This work analyzes the challenges, benefits, and drawbacks of the different execution modes available for TCN-based KWS inference on dedicated hardware. With this objective, this research contributes to: 1) presenting a complete deep learning accelerator optimized for TCN inference; 2) evaluating the impact on performance and power of the different deployment options for TCN inference applied to KWS obtaining up to 8 $\mu \text{W}$ for real-time operation; and 3) optimizing real-time power consumption for KWS inference by exploiting the use of cascaded neural networks (NNs), achieving up to 35% additional power savings. [ABSTRACT FROM AUTHOR]