Abstract Neural architecture search can be used to find convolutional neural architectures that are precise and robust while enjoying enough speed for industrial image processing applications. In this paper, our goal is to achieve optimal convolutional neural networks (CNNs) for multiple‐templates matching for applications such as licence plates detection (LPD). We perform an iterative local neural architecture search for the models with minimum validation error as well as low computational cost from our search space of about 32 billion models. We describe the findings of the experience and discuss the specifications of the final optimal architectures. About 20‐times error reduction and 6‐times computational complexity reduction is achieved over our engineered neural architecture after about 500 neural architecture evaluation (in about 10 h). The typical speed of our final model is comparable to classic template matching algorithms while performing more robust and multiple‐template matching with different scales.

Vehicle speed is known as one of the most important parameters in the various goals for driving. Although the most critical goal of driving is safety, each driver can consider another secondary goal, such as reducing travel time, economical driving, green driving, passenger comfort, etc. In this paper, we propose a smartphone-based application that utilizes the cloud/fog computing service infrastructure in order to recommend speed according to primary safety goal and another secondary objective. Due to the strategic nature of the driving speed selection issue, we have modeled the problem as a game that the drivers are the players, and the speed of the vehicle is their strategy. We solve the proposed game via evolutionary dynamics with appropriate convergence time, and the resulting equilibrium profile is announced to the drivers as the recommended speed at specified time intervals. Finally, the experimental results obtained from the simulation of the proposed scheme confirm that the deviation from the proposed equilibrium velocity is not profitable for the offending player. Indeed, deviation from equilibrium conditions leads to the negative impacts on the formal evaluation parameters presented in this paper.

Automatic human activity recognition is an integral part of any interactive application involving humans (e.g., human-robot interaction systems). One of the main challenges for activity recognition is the diversity in the way individuals often perform activities. Furthermore, changes in any of the environment factors (i.e., illumination, complex background, human body shapes, viewpoint, etc.) intensify this challenge. In addition, there are different types of activities that robots need to interpret for seamless interaction with humans. Some activities are short, quick, and simple (e.g., sitting), while others may be detailed/complex, and spread throughout a long span of time (e.g., washing mouth). In this article, we recognize the activities within the context of graphical models in a sequence-labeling framework based on skeleton data. We propose a new structured prediction strategy based on probabilistic graphical models (PGMs) to recognize both types of activities (i.e., complex and simple). These activity types are often spanned in very diverse subspaces in the space of all possible activities, which would require different model parameterizations. In order to deal with these parameterization and structural breaks across models, a category-switching scheme is proposed to switch over the models based on the activity types. For parameter optimization, we utilize a distributed structured prediction technique to implement our model in a distributed setting. The method is tested on three widely used datasets (CAD-60, UT-Kinect, and Florence 3-D) that cover both activity types. The results illustrate that our proposed method is able to recognize simple and complex activities while the previous work concentrated on only one of these two main types.

One-class classification (OCC) poses as an essential component in many machine learning and computer vision applications, including novelty, anomaly, and outlier detection systems. With a known definition for a target or normal set of data, one-class classifiers can determine if any given new sample spans within the distribution of the target class. Solving for this task in a general setting is particularly very challenging, due to the high diversity of samples from the target class and the absence of any supervising signal over the novelty (nontarget) concept, which makes designing end-to-end models unattainable. In this article, we propose an adversarial training approach to detect out-of-distribution samples in an end-to-end trainable deep model. To this end, we jointly train two deep neural networks, R and D . The latter plays as the discriminator while the former, during training, helps D characterize a probability distribution for the target class by creating adversarial examples and, during testing, collaborates with it to detect novelties. Using our OCC, we first test outlier detection on two image data sets, Modified National Institute of Standards and Technology (MNIST) and Caltech-256. Then, several experiments for video anomaly detection are performed on University of Minnesota (UMN) and University of California, San Diego (UCSD) data sets. Our proposed method can successfully learn the target class underlying distribution and outperforms other approaches.

Novelty detection is the process of identifying the observation(s) that differ in some respect from the training observations (the target class). In reality, the novelty class is often absent during training, poorly sampled or not well defined. Therefore, one-class classifiers can efficiently model such problems. However, due to the unavailability of data from the novelty class, training an end-to-end deep network is a cumbersome task. In this paper, inspired by the success of generative adversarial networks for training deep models in unsupervised and semi-supervised settings, we propose an end-to-end architecture for one-class classification. Our architecture is composed of two deep networks, each of which trained by competing with each other while collaborating to understand the underlying concept in the target class, and then classify the testing samples. One network works as the novelty detector, while the other supports it by enhancing the inlier samples and distorting the outliers. The intuition is that the separability of the enhanced inliers and distorted outliers is much better than deciding on the original samples. The proposed framework applies to different related applications of anomaly and outlier detection in images and videos. The results on MNIST and Caltech-256 image datasets, along with the challenging UCSD Ped2 dataset for video anomaly detection illustrate that our proposed method learns the target class effectively and is superior to the baseline and state-of-the-art methods.

The 3D design of SRAM LLCs has made the thermal problem, even more, sever and therefore incurs more leakage energy consumption compared to conventional SRAM cache architectures in 2Ds due to dense integration. In this paper, we propose a runtime cache architecture called NIZCache. The core idea of NIZCache is to use the non-uniform distribution of the accesses, invalid lines, and zero-value lines in banks of last level cache for decreasing both dynamic and leakage energy. The proposed architecture that is based on non-uniform cache architectures (NUCA) disables cache banks that have low accesses and high invalid and zero lines and leads to high energy-efficiency. The Experimental results show that the proposed method improves the energy-delay product by about 17% on average under PARSEC benchmarks compared to previous work.

MicroRNAs (miRNAs) are short non-coding RNAs which bind to mRNAs and regulate their expression. MiRNAs have been found to be associated with initiation and progression of many complex diseases. Investigating miRNAs and their targets can thus help develop new therapies by designing anti-miRNA oligonucleotides. While existing computational approaches can predict miRNA targets, these predictions have low accuracy. In this paper, we propose a two-step approach to refine the results of sequence-based prediction algorithms. The first step, which is based on our previous work, uses an ensemble learning approach that combines multiple existing methods. The second step utilizes support vector machine (SVM) classifiers in one- and two-class modes to infer miRNA-mRNA interactions based on both binding features, as well as network features extracted from gene regulatory network. Experimental results using two real data sets from TCGA indicate that the use of two-class SVM classification significantly improves the precision of miRNA-mRNA prediction.

In this article, we present a convex optimization model to design a stacked hybrid memory system contains eDRAM and STT-RAM banks with minimum write energy consumption of STT-RAM memory banks and minimum refresh energy of eDRAM banks with efficient number of TSVs in embedded CMP. Our convex model optimizes numbers and placement of memory banks from different technologies on the memory layer and finding optimal number and optimal placement of TSVs while satisfy maximum Load on TSVs. Experimental results on PARSEC benchmark show that the proposed Architecture improves energy-delay product (EDP) by 51% on average.

One of the main abilities that the robots need to maintain is to efficiently communicate with people in a humanly manner. Thus, human activity recognition (HAR) would be an integral part of such a human-robot interaction system. One of the major challenges in HAR is that the individuals perform their activities in different manners. Furthermore, there is a very wide range of different types of activities that the robots would require to understand. Some activities are simple, quick and short (e.g., sit down), while many others are complex, have many details and span through a long range of time (e.g., wearing contact lens). In this paper, we model the recognition of activities into a sequence-labeling problem and propose a new probabilistic graphical model (PGM) that can recognize both short/long-range activities, by introducing a hierarchical classification model and including extra links and loopy conditions in our PGM. To optimize the PGM and obtain its parameters during training, we use a structured prediction technique, a general framework that involves latent structured support vector machines (LSSVM) and hidden-state conditional random fields (HCRF). We evaluate our method on two widely used datasets (CAD-60 & UT-Kinect) that contain both activity types. Our obtained results are promising and show that our method can recognize both types of activities effectively, while most of the previous works only focused on one of these two major types. We further explore distributed processing techniques, since our method can easily be distributed over processing nodes. We also propose an efficient divide-and-merge technique to further speedup the training step.

This paper proposes a fast and reliable method for anomaly detection and localization in video data showing crowded scenes. Time-efficient anomaly localization is an ongoing challenge and subject of this paper. We propose a cubicpatch- based method, characterised by a cascade of classifiers, which makes use of an advanced feature-learning approach. Our cascade of classifiers has two main stages. First, a light but deep 3D auto-encoder is used for early identification of "many" normal cubic patches. This deep network operates on small cubic patches as being the first stage, before carefully resizing remaining candidates of interest, and evaluating those at the second stage using a more complex and deeper 3D convolutional neural network (CNN). We divide the deep autoencoder and the CNN into multiple sub-stages which operate as cascaded classifiers. Shallow layers of the cascaded deep networks (designed as Gaussian classifiers, acting as weak single-class classifiers) detect "simple" normal patches such as background patches, and more complex normal patches are detected at deeper layers. It is shown that the proposed novel technique (a cascade of two cascaded classifiers) performs comparable to current top-performing detection and localization methods on standard benchmarks, but outperforms those in general with respect to required computation time.

In this article, we present a convex optimization model to design a stacked hybrid memory system to improve performance and reduce energy consumption of the chip-multiprocessor (CMP). Our convex model optimizes numbers and placement of SRAM and STT-RAM memories on the memory layer, and efficiently maps applications/threads on cores in the core layer. Power consumption that is the main challenge in the dark silicon era is represented as a power constraint in this work and it is satisfied by the detailed optimization model in order to design a dark silicon aware 3D CMP. Experimental results show that the proposed architecture considerably improves the energy-delay product (EDP) and performance of the 3D CMP compared to the Baseline memory design.

Technology scaling has enabled increasing number of cores on a chip in Chip-Multiprocessors (CMPs). As the number of cores increases, the overall system will need to provide more cache resources to feed all the cores. However, increasing the size of each cache level in the cache hierarchy of CMPs mitigates the large off-chip memory access latencies and bandwidth constraints. Moreover, cache hierarchy is known as one of the most power-hungry components in many-core CMPs because leakage power within the cache systems has become a significant contributor in the overall chip power budget in deep sub-micron as well as dark silicon era. Due to the many advantages of Non-Volatile Memory (NVM) technology such as high density, near zero leakage, and non-volatility, in this paper, we focus on exploiting such memories in the cache hierarchy. Specifically, we focus on 3D CMPs to decrease the leakage power consumption and mitigating the dark silicon phenomenon. Experimental results show that the proposed method on average improves the throughput by 45.5% and energy-delay product by 56% when compared to the conventional single cache technology.

Large-scale multi-label classification has always been of great interest for researchers. The difficulty with such problems is the huge amount of data that should be processed, possibly in multiple paths. This amount of data does not fit in the memory of a single computer and that is the bottle-nec k for many large-scale applications. On the other hand, matrix completion is a great tool for many applications, including classification. It is a great tool for modeling the data and finding the outliers and noises within the data. In this paper, we develop a distributed matrix completion method for multi-label classification. To do this, we first propose a simple distributed algorithm for minimizing the nuclear norm of a matrix to recover its low-rank representation, which is then generalized for the classification problem. Several synthetic and real datasets are used to verify both the distributed nuclear norm minimization and the distributed matrix completion approach. The results indicate that the proposed algorithm outperforms state-of-the-art methods for large-scale classification.

Bag of visual words model has recently attracted much attention from computer vision society because of its notable success in analysing images and exploring their content. This study improves this model by utilising the adjacency information between words. To explore this information, a binary tree structure is constructed from the visual words in order to model the is – a relationships in the vocabulary. Informative nodes of this tree are extracted by using the χ 2 criterion and are used to capture the adjacency information of visual words. This approach is a simple and computationally effective way for modelling the spatial relations of visual words, which improves the image classification performance. The authors evaluated our method for visual classification of three known datasets: 15 natural scenes, Caltech‐101 and Graz‐01.

In this paper, an Iranian vehicle license plate recognition system based on a new localization approach, which is modified to reflect the local context, is proposed, along with a hybrid classifier that recognizes license plate characters. The method presented here is based on a modified template-matching technique by the analysis of target color pixels to detect the location of a vehicle’s license plate. A modified strip search enables localization of the standard color-geometric template utilized in Iran and several European countries. This approach uses periodic strip search to find the hue of each pixel on demand. In addition, when a group of target pixels is detected, it is analyzed to verify that its shape and aspect ratio match those of the standard license plate. In addition to being scale and rotation invariant, this method avoids time-consuming image algorithms and transformations for the whole image pixels, such as resizing and Hough, Fourier, and wavelet transforms, thereby cutting down the detection response time. License plate characters are recognized by a hybrid classifier that comprises a decision tree and a support vector machine with a homogeneous fifth-degree polynomial kernel. The performance detection rate and the overall system performance achieved are 96% and 94%, respectively.

Three-dimensional network-on-chips are suitable communication fabrics for high-density 3D many-core ICs. Such networks have shorter communication hop count, compared to 2D NoCs, and enjoy fast and power efficient TSV wires in vertical links. Unfortunately, the fabrication process of TSV connections has not matured yet, which results in poor vertical links yield. In this work, we address this challenge and introduce AFRA, a deadlock-free routing algorithm for 3D mesh-based NoCs that tolerates faults on vertical links. AFRA is designed to be simple, high performance, and robust. The simplicity is achieved by applying ZXY and XZXY routings in the absence and presence of fault, respectively. Furthermore, AFRA, as will be proved, is deadlock-free when all vertical faulty links have the same direction. This enables the routing to save virtual channels for performance rather than scarifying them for deadlock avoidance. Finally, AFRA provides robustness, which means supporting connection for all possible pairs of communicating nodes in high fault rates. AFRA is evaluated, though cycle accurate network simulation, and is compared with planar adaptive routing. Results reveal that AFRA significantly outperforms planar adaptive routing in both synthetic and real traffic patterns. In addition, the robustness of AFRA is calculated analytically.

We analyze the effect of using a vehicular ad-hoc network (VANET) on accident avoidance. As shown in our analysis, a higher frequency of safety packets can prevent accidents, even for high speed vehicles and dense roads. To overcome connectivity problems in blind crossing situations, a genetic algorithm (GA) based method is presented for VANET infrastructure planning. The proposed approach tries to remove coverage sight holes in low sight distance cases in a traveling path in the road. In such places, drivers might not have enough sight for proper action and also environmental obstacles prevent direct communication between vehicles. Furthermore, curved roads affect mobility. Simulation results show that the density of vehicles is increased right before a curve and is decreased after that. Therefore, in this kind of road, a high frequency of packet generation may not act well in accident avoidance. The method proposed in this paper tries to cover such places considering the lowest safety distances according to traffic theory. For this, the road must be covered directly by infrastructure. Therefore, the problem is to find the best number and also positions of road side units. Using GA, the algorithm minimizes the summation of total uncovered and overlapped points in the roads which are covered by more than one antenna. Simulation on a real road map confirmed the capabilities of the proposed approach.

A critical issue of Neural Network based large-scale data mining algorithms is how to speed up their learning algorithm. This problem is particularly challenging for Error Back-Propagation (EBP) algorithm in Multi-Layered Perceptron (MLP) Neural Networks due to their significant applications in many scientific and engineering problems. In this paper, we propose an Adaptive Variable Learning Rate EBP algorithm to attack the challenging problem of reducing the convergence time in an EBP algorithm, aiming to have a high-speed convergence in comparison with standard EBP algorithm. The idea is inspired from adaptive filtering, which leaded us into two semi-similar methods of calculating the learning rate. Mathematical analysis of AVLR-EBP algorithm confirms its convergence property. The AVLR-EBP algorithm is utilized for data classification applications. Simulation results on many well-known data sets shall demonstrate that this algorithm reaches to a considerable reduction in convergence time in comparison to the standard EBP algorithm. The proposed algorithm, in classifying the IRIS, Wine, Breast Cancer, Semeion and SPECT Heart datasets shows a reduction of the learning epochs relative to the standard EBP algorithm.

The IEEE 802.16 standard of wireless mesh networks includes various scheduling algorithms, both centralized and distributed, to determine the proper time slot that each wireless node can transmit its data. In this paper, we introduce a new scheduling scheme to improve scalability and resource utilization in wireless mesh networks where all nodes compete to gain access to the transmission time slot. In our proposed method, the transmission time of a node is modified according to the transmission status of its neighbors, resulting in a shorter holdoff time, and consequently, reducing the transmission delay and throughput. The simulation results confirm the efficiency and performance of the proposed algorithm over the IEEE 802.16-mesh standard in terms of throughput and delay.

With reducing feature size of transistors and increasing number of cores on a single chip, fault tolerance and reliability have become two significant challenges for IC designers. Since chip design is extremely cost-sensitive, the fault tolerance redundancy must be provided at a reasonable cost. In this paper, a fault tolerant NoC architecture with cores linked to two switches instead of one, is proposed. This architecture is able to save cores with a faulty switch. Also, to be more efficient and to compensate this redundancy, a new routing algorithm is suggested that can be dynamically reconfigured to escape faulty switches. According to evaluation of the proposed architecture, latency and reliability are improved.

VANET is new generation of ad hoc networks that implement between vehicles on a road. Because of high mobility, routing in VANET has more problems than MANET. Thereby, in this paper we propose a modification on AODV as MANET routing protocol to make it adaptive for VANET. When a node is mobile, it has three mobility parameters: position, direction and speed. In our method, we have used direction as most important parameter to select next hop during a route discovery phase. With respect to mobility model, if nodes has same direction with source and/or destination nodes, our solution might selects them as a next hop. Position is another parameter that we used for next hop selection.

Background estimation is one of the most challenging phases in extracting foreground objects from video sequences. In this paper we present a background modeling approach that uses the similarity of frames to extract background areas from the video sequence. We use a window over the frames history and compute the similarity between the selected frames of this window as a similarity window. The properties of similarity window depend on the characteristics of the scene and can be adjusted parametrically. Our primary results show that if proper parameters are chosen, this method can give a good approximation of the background model.

One of the key issues in the design of multiprocessors system-on-chip (MP-SoCs), multicomputers, and peer-to-peer networks is the development of an efficient communication network to provide high throughput and low latency and its ability to survive beyond the failure of individual components. Generally, the faulty components may be coalesced into fault regions, which are classified into convex and concave shapes. In this paper, we propose a mathematical solution for counting the number of common fault patterns in a 2-D mesh interconnect network including both convex (I-shape, II-shape, square-shape) and concave (L-shape, U- shape, T-shape, +-shape, H-shape) regions. The results presented in this paper which have been validated through simulation experiments can play a key role when studying, particularly, the performance analysis of fault-tolerant routing algorithms and measure of a network fault-tolerance expressed as the probability of a disconnection.

This paper presents a new algorithm for background subtraction that can model the background image from a sequence of images, even if there are foreground objects in each image frame. In contrast with Gaussian mixture model algorithm, in our proposed method the problem of distinguishing between background and foreground kernels becomes trivial. The key idea of our method lies in the identification of the background based on QR-decomposition method in linear algebra. R-values taken from QR-decomposition can be applied to decompose a given system to indicate the degree of the significance of the decomposed parts. We split the image into small blocks and select the background blocks with the weakest contribution, according to the assigned R-values. Simulation results show the better background detection performance with respect to some others.

One of the fundamental problems in parallel computing is how to efficiently perform routing in a faulty network each component of which fails with some probability. This paper presents a comparative performance study of ten prominent adaptive fault-tolerant routing algorithms in wormhole-switched 2D mesh interconnect networks. These networks carry a routing scheme suggested by Boppana and Chalasani as an instance of a fault-tolerant method. The suggested scheme is widely used in the literature to achieve high adaptivity and support inter-processor communications in parallel computer systems due to its ability to preserve both communication performance and fault-tolerant demands in these networks. The performance measures studied are the throughput, average message latency and average usage of virtual channels per node. Results obtained through simulation suggest two classes of presented routing schemes as high performance candidate in most faulty networks.

Massively parallel computing systems are being built with hundreds or thousands of components such as nodes, links, memories, and connectors. The failure of a component in such systems will not only reduce the computational power but also alter the network's topology. The software-based fault-tolerant routing algorithm is a popular routing to achieve fault-tolerance capability in networks. This algorithm is initially proposed only for two dimensional networks (Suh et al., 2000). Since, higher dimensional networks have been widely employed in many contemporary massively parallel systems; this paper proposes an approach to extend this routing scheme to these indispensable higher dimensional networks. Deadlock and livelock freedom and the performance of presented algorithm, have been investigated for networks with different dimensionality and various fault regions. Furthermore, performance results have been presented through simulation experiments