Real-time matting is a challenging research in deep learning. Conventional CNN (Convolutional Neural Networks) approaches are easy to misjudge the foreground and background semantic and have blurry matting edges, which result from CNN’s limited concentration on global context due to receptive field. We propose a real-time matting approach called RMViT (Real-time matting with Vision Transformer) with Transformer structure, attention and content-aware guidance to solve issues above. The semantic accuracy improves a lot due to the establishment of global context and long-range pixel information. The experiments show our approach exceeds a 30% reduction in error metrics compared with existing real-time matting approaches.

With maturation of 5G technology in recent years, multimedia services such as live video streaming and online games on the Internet have flourished. These multimedia services frequently require low latency, which pose a significant challenge to compute the high latency requirements multimedia tasks. Mobile edge computing (MEC), is considered a key technology solution to address the above challenges. It offloads computation-intensive tasks to edge servers by sinking mobile nodes, which reduces task execution latency and relieves computing pressure on multimedia devices. In order to use MEC paradigm reasonably and efficiently, resource allocation has become a new challenge. In this paper, we focus on the multimedia tasks which need to be uploaded and processed in the network. We set the optimization problem with the goal of minimizing the latency and energy consumption required to perform tasks in multimedia devices. To solve the complex and non-convex problem, we formulate the optimization problem as a distributed deep reinforcement learning (DRL) problem and propose a federated Dueling deep Q-network (DDQN) based multimedia task offloading and resource allocation algorithm (FDRL-DDQN). In the algorithm, DRL is trained on the local device, while federated learning (FL) is responsible for aggregating and updating the parameters from the trained local models. Further, in order to solve the not identically and independently distributed (non-IID) data problem of multimedia devices, we develop a method for selecting participating federated devices. The simulation results show that the FDRL-DDQN algorithm can reduce the total cost by 31.3% compared to the DQN algorithm when the task data is 1000 kbit, and the maximum reduction can be 35.3% compared to the traditional baseline algorithm.

In this study, we introduce a reservoir-based one-dimensional (1D) convolutional neural network that processes time-series data at a low computational cost, and investigate its performance and training time. Experimental results show that the proposed network consumes lower training computational costs and that it outperforms the conventional reservoir computing in a sound-classification task.

Crowdsourcing becomes a popular data-collection method to relieve the burden of high cost and latency for data-gathering. Since the involved users in crowdsourcing are volunteers, need incentives to encourage them to provide data. However, the current incentive mechanisms mostly pay attention to the data quantity, while ignoring the data quality. In this paper, we design a Data-quality awaRe IncentiVe mEchanism (DRIVE) for collaborative tasks based on the Stackelberg game to motivate users with high quality, the highlight of which is the dynamic reward allocation scheme based on the proposed data quality evaluation method. In order to guarantee the data quality evaluation response in real-time, we introduce the mobile edge computing framework. Finally, one case study is given and its real-data experiments demonstrate the superior performance of DRIVE.

The segmentation performance of fresh tea sprouts is inadequate due to the uncontrollable posture. A novel method for Fresh Tea Sprouts Segmentation based on Capsule Network (FTS-SegCaps) is proposed in this paper. The spatial relationship between local parts and whole tea sprout is retained and effectively utilized by a deep encoder-decoder capsule network, which can reduce the effect of tea sprouts with uncontrollable posture. Meanwhile, a patch-based local dynamic routing algorithm is also proposed to solve the parameter explosion problem. The experimental results indicate that the segmented tea sprouts via FTS-SegCaps are almost coincident with the ground truth, and also show that the proposed method has a better performance than the state-of-the-art methods.

Owing to the COVID-19 pandemic, many academic conferences are now being held online. Our study focuses on online video conferences, where participants can watch pre-recorded embedded videos on a conference website. In online video conferences, participants must efficiently find videos that match their interests among many candidates. There are few opportunities to encounter videos that they may not have planned to watch but may be of interest to them unless participants actively visit the conference. To alleviate these problems, the introduction of a recommender system seems promising. In this paper, we implemented typical recommender systems for the online video conference with 4,000 participants and analyzed users’ behavior through A/B testing. Our results showed that users receiving recommendations based on collaborative filtering had a higher continuous video-viewing rate and spent longer on the website than those without recommendations. In addition, these users were exposed to broader videos and tended to view more from categories that are usually less likely to view together. Furthermore, the impact of the recommender system was most significant among users who spent less time on the site.

Forecasting time-series data is useful in many fields, such as stock price predicting system, autonomous driving system, weather forecast, etc. Many existing forecasting models tend to work well when forecasting short-sequence time series. However, when working with long sequence time series, the performance suffers significantly. Recently, there has been more intense research in this direction, and Informer is currently the most efficient predicting model. Informer’s main drawback is that it does not allow for incremental learning. In this paper, we propose a Fast Informer called Finformer, which addresses the above bottleneck by reducing the training/predicting time of Informer. Finformer can efficiently compute the positional/temporal/value embedding and Query/Key/Value of the self-attention incrementally. Theoretically, Finformer can improve the speed of both training and predicting over the state-of-the-art model Informer. Extensive experiments show that Finformer is about 26% faster than Informer for both short and long sequence time series prediction. In addition, Finformer is about 20% faster than InTrans for the general Conv1d, which is one of our previous works and is the predecessor of Finformer.

This Letter focuses on deep learning-based monkeys' head swing counting problem. Nowadays, there are very few papers on monkey detection, and even fewer papers on monkeys' head swing counting. This research tries to fill in the gap and try to calculate the head swing frequency of monkeys through deep learning, where we further extend the traditional target detection algorithm. After analyzing object detection results, we localize the monkey's actions over a period. This Letter analyzes the task of counting monkeys' head swings, and proposes the standard that accurately describes a monkey's head swing. Under the guidance of this standard, the monkeys' head swing counting accuracy in 50 test videos reaches 94.23%.

With the recent advances in IoT, there is a growing interest in multi-surface computing, where a mobile app can cooperatively utilize multiple devices' surfaces. We propose a novel framework that seamlessly augments mobile apps with multi-surface computing capabilities. It enables various apps to employ multiple surfaces with acceptable performance.

Cloud computing has become the mainstream computing paradigm nowadays. More and more data owners (DO) choose to outsource their data to a cloud service provider (CSP), who is responsible for data management and query processing on behalf of DO, so as to cut down operational costs for the DO.  However, in real-world applications, CSP may be untrusted, hence it is necessary to authenticate the query result returned from the CSP.  In this paper, we consider the problem of approximate string query result authentication in the context of database outsourcing. Based on Merkle Hash Tree (MHT) and Trie, we propose an authenticated tree structure named MTrie for authenticating approximate string query results. We design efficient algorithms for query processing and query result authentication. To verify effectiveness of our method, we have conducted extensive experiments on real datasets and the results show that our proposed method can effectively authenticate approximate string query results.

In this paper, we investigated the overfitting characteristics of nonlinear equalizers based on an artificial neural network (ANN) and the Volterra series transfer function (VSTF), which were designed to compensate for optical nonlinear waveform distortion in optical fiber communication systems. Linear waveform distortion caused by, e.g., chromatic dispersion (CD) is commonly compensated by linear equalizers using digital signal processing (DSP) in digital coherent receivers. However, mitigation of nonlinear waveform distortion is considered to be one of the next important issues. An ANN-based nonlinear equalizer is one possible candidate for solving this problem. However, the risk of overfitting of ANNs is one obstacle in using the technology in practical applications. We evaluated and compared the overfitting of ANN- and conventional VSTF-based nonlinear equalizers used to compensate for optical nonlinear distortion. The equalizers were trained on repeated random bit sequences (RRBSs), while varying the length of the bit sequences. When the number of hidden-layer units of the ANN was as large as 100 or 1000, the overfitting characteristics were comparable to those of the VSTF. However, when the number of hidden-layer units was 10, which is usually enough to compensate for optical nonlinear distortion, the overfitting was weaker than that of the VSTF. Furthermore, we confirmed that even commonly used finite impulse response (FIR) filters showed overfitting to the RRBS when the length of the RRBS was equal to or shorter than the length of the tapped delay line of the filters. Conversely, when the RRBS used for the training was sufficiently longer than the tapped delay line, the overfitting could be suppressed, even when using an ANN-based nonlinear equalizer with 10 hidden-layer units.

Given an odd prime q and an integer m ≤ q, a binary mq × q2 quasi-cyclic parity-check matrix H(m, q) can be constructed for an array low-density parity-check (LDPC) code C (m, q). In this letter, we investigate the first separating redundancy of C (m, q). We prove that H (m, q) is 1-separating for any pair of (m, q), from which we conclude that the first separating redundancy of C (m, q) is upper bounded by mq. Then we show that our upper bound on the first separating redundancy of C (m, q) is tighter than the general deterministic and constructive upper bounds in the literature. For m=2, we further prove that the first separating redundancy of C (2, q) is 2q for any odd prime q. For m ≥ 3, we conjecture that the first separating redundancy of C (m, q) is mq for any fixed m and sufficiently large q.

We considered the problem of forecasting the degradation recovery process of civil structures for prognosis and health management. In this process, structural health degrades over time but recovers when a maintenance intervention is performed. Maintenance interventions are typically recorded in terms of date and type. Such records can be represented as binary time series. Using binary maintenance intervention records, we forecast the process by using Long Short-Term Memory (LSTM). In this study, we experimentally examined how to feed binary time series data into LSTM. To this end, we compared the concatenation and reinitialization methods. The former is used to concatenate maintenance intervention records and health data and feed them into LSTM. The latter is used to reinitialize the LSTM internal memory when maintenance intervention is performed. The experimental results with the synthetic data revealed that the concatenation method outperformed the reinitialization method.

ARM chips with low energy consumption and low-cost investment have been rapidly applied to smart office and smart entertainment including cloud mobile phones and cloud games. This paper first summarizes key technologies and development status of the above scenarios including CPU, memory, IO hardware virtualization characteristics, ARM hypervisor and container, GPU virtualization, network virtualization, resource management and remote transmission technologies. Then, in view of the current lack of publicly referenced ARM cloud constructing solutions, this paper proposes and constructs an implementation framework for building an ARM cloud, and successively focuses on the formal definition of virtualization framework, Android container system and resource quota management methods, GPU virtualization based on API remoting and GPU pass-through, and the remote transmission technology. Finally, the experimental results show that the proposed model and corresponding component implementation methods are effective, especially, the pass-through mode for virtualizing GPU resources has higher performance and higher parallelism.

Approximate computing (AC) saves energy and improves performance by introducing approximation into computation in error-torrent applications. This work focuses on an AC strategy that accurately performs important computations and approximates others. In order to make AC circuits practical, we need to determine which computation is how important carefully, and thus need to appropriately approximate the redundant computation for maintaining the required computational quality. In this paper, we focus on the importance of computations at the flip-flop (FF) level and propose a novel importance evaluation methodology. The key idea of the proposed methodology is a two-step fault injection algorithm to extract the near-optimal set of redundant FFs in the circuit. In the first step, the proposed methodology performs the FI simulation for each FF and extracts the candidates of redundant FFs. Then, in the second step, the proposed methodology extracts the set of redundant FFs in a binary search manner. Thanks to the two-step strategy, the proposed algorithm reduces the complexity of architecture exploration from an exponential order to a linear order without understanding the functionality and behavior of the target application program. Experimental results show that the proposed methodology identifies the candidates of redundant FFs depending on the given constraints. In a case study of an image processing accelerator, the truncation for identified redundant FFs reduces the circuit area by 29.6% and saves power dissipation by 44.8% under the ASIC implementation while satisfying the PSNR constraint. Similarly, the dynamic power dissipation is saved by 47.2% under the FPGA implementation.

Distributed computing is one of the powerful solutions for computational tasks that need the massive size of dataset. Lagrange coded computing (LCC), proposed by Yu et al. [15], realizes private and secure distributed computing under the existence of stragglers, malicious workers, and colluding workers by using an encoding polynomial. Since the encoding polynomial depends on a dataset, it must be updated every arrival of new dataset. Therefore, it is necessary to employ efficient algorithm to construct the encoding polynomial. In this paper, we propose Newton coded computing (NCC) which is based on Newton interpolation to construct the encoding polynomial. Let K, L, and T be the number of data, the length of each data, and the number of colluding workers, respectively. Then, the computational complexity for construction of an encoding polynomial is improved from O(L(K+T)log 2(K+T)log log (K+T)) for LCC to O(L(K+T)log (K+T)) for the proposed method. Furthermore, by applying the proposed method, the computational complexity for updating the encoding polynomial is improved from O(L(K+T)log 2(K+T)log log (K+T)) for LCC to O(L) for the proposed method.

In the field of machine learning security, as one of the attack surfaces especially for edge devices, the application of side-channel analysis such as correlation power/electromagnetic analysis (CPA/CEMA) is expanding. Aiming to evaluate the leakage resistance of neural network (NN) model parameters, i.e. weights and biases, we conducted a feasibility study of CPA/CEMA on floating-point (FP) operations, which are the basic operations of NNs. This paper proposes approaches to recover weights and biases using CPA/CEMA on multiplication and addition operations, respectively. It is essential to take into account the characteristics of the IEEE 754 representation in order to realize the recovery with high precision and efficiency. We show that CPA/CEMA on FP operations requires different approaches than traditional CPA/CEMA on cryptographic implementations such as the AES.

In this paper, we study BALL COLLECTING WITH LIMITED ENERGY, which is a problem of scheduling robots with limited energy confined to a line to catch moving balls that eventually cross the line. For this problem, we show the NP-completeness of the general case and some algorithmic results for some cases with a small number of robots.

Following the formulation of Support Vector Regression (SVR), we consider a regression analogue of soft margin optimization over the feature space indexed by a hypothesis class H. More specifically, the problem is to find a linear model w ∈ ℝH that minimizes the sum of ρ-insensitive losses over all training data for as small ρ as posssible, where the ρ-insensitive loss for a single data (xi, yi) is defined as max{|yi - ∑h whh(xi)| - ρ, 0}. Intuitively, the parameter ρ and the ρ-insensitive loss are defined analogously to the target margin and the hinge loss in soft margin optimization, respectively. The difference of our formulation from SVR is two-fold: (1) we consider L1-norm regularization instead of L2-norm regularization, and (2) the feature space is implicitly defined by a hypothesis class instead of a kernel. We propose a boosting-type algorithm for solving the problem with a theoretically guaranteed convergence rate under a natural assumption on the weak learnability.

A T-junction orthomode transducer (OMT) is a waveguide component that separates two orthogonal linear polarizations in the same frequency band. It has a common circular waveguide short-circuited at one end and two branch rectangular waveguides arranged in opposite directions near the short circuit. One of the advantages of a T-junction OMT is its short axial length. However, the two rectangular ports, which need to be orthogonal, have different levels of performance because of asymmetry. We therefore propose a uniaxially symmetrical T-junction OMT, which is configured such that the two branch waveguides are tilted 45° to the short circuit. The uniaxially symmetrical configuration enables same levels of performance for the two ports, and its impedance matching is easier compared to that for the conventional configuration. The polarization separation principle can be explained using the principles of orthomode junction (OMJ) and turnstile OMT. Based on calculations, the proposed configuration demonstrated a return loss of 25dB, XPD of 30dB, isolation of 21dB between the two branch ports, and loss of 0.25dB, with a bandwidth of 15% in the K band. The OMT was then fabricated as a single piece via 3D printing and evaluated against the calculated performance indices.