In this study, we construct two kinds of data sets from distinct time periods, both comprising line-of-sight magnetograms and knowledge-informed features. We develop eight models for forecasting ≥M-class flares within 24 hr, including the image-based convolutional neural network (CNN), CNN-BiLSTM, CNN-BiLSTM-Attention, and Vision Transformer models, as well as the knowledge-informed neural network, BiLSTM, BiLSTM-Attention, and iTransformer models. We analyze the importance of knowledge-informed features by assessing categorical and probabilistic performance using the true skill statistic (TSS) and the Brier skill score (BSS), respectively. This is the first time the iTransformer has been applied to flare forecasting. Subsequently, we compare the forecasting performance of the eight models. Then, we investigate the generalization ability of the models across three different data products. Finally, we fairly compare the forecasting performance of iTransformer with that of the currently advanced NASA/CCMC models. The major results are as follows. (1) The R_VALUE feature consistently shows the best performance in both categorical and probabilistic forecasting for the knowledge-informed models. (2) The iTransformer yields the highest forecasting performance, with TSS and BSS scores of 0.768 ± 0.072 and 0.513 ± 0.063, respectively. The knowledge-informed deep learning models consistently outperform image-based models. (3) The three image-based models demonstrate good generalization performance in categorical forecasting on Space-weather Helioseismic and Magnetic Imager (HMI) Active Region Patches (SHARP), HMI, and Full-Disk Magnetograph (FMG), while the two knowledge-informed models exhibit excellent generalization performance on SHARP and HMI. This is the first time that FMG magnetograms and knowledge-informed features are used for flare forecasting. Additionally, the five models also demonstrate strong generalization ability on SHARP across different time periods. (4) The iTransformer exhibits superior forecasting performance compared to NASA/CCMC.

针对传统预测模型因分析铝厂时序数据时历史数据量大而无法快速挖掘实时数据隐含的知识信息,导致预测效率低的问题,提出一种基于生长神经气改进模糊神经网络(GNG-ANFIS)全局高效的时序混合预测模型.该模型首先利用生长神经气动态跟踪采集到的时序数据来识别数据奇异点,进而筛选有效数据;然后利用改进后的黑猩猩算法对传统模糊神经网络进行优化;最后,结合铝电解生产过程中铝液杂质铁含量时序数据验证该模型的性能.实验结果表明,混合模型在减少训练时间的情况下仍能准确预测铁含量时序数据,验证了其可行性.

The cyclin-dependent protein kinases (CDKs) are protein-serine/threonine kinases with crucial effects on the regulation of cell cycle and transcription. CDKs can be a hallmark of cancer since their excessive expression could lead to impaired cell proliferation. However, the selectivity profile of most developed CDK inhibitors is not enough, which have hindered the therapeutic use of CDK inhibitors. In this study, we propose a multitask deep learning framework called BiLAT based on SMILES representation for the prediction of the inhibitory activity of molecules on eight CDK subtypes (CDK1, 2, 4-9). The framework is mainly composed of an improved bidirectional long short-term memory module BiLSTM and the encode layer of the Transformer framework. Additionally, the data enhancement method of SMILES enumeration is applied to improve the performance of the model. Compared with baseline predictive models based on three conventional machine learning methods and two multitask deep learning algorithms, BiLAT achieves the best performance with the highest average AUC, ACC, F1-score, and MCC values of 0.938, 0.894, 0.911, and 0.715 for the test set. Moreover, we constructed a targeted external data set CDK-Dec for the CDK family, which mainly contains bait values screened by 3D similarity with active compounds. This dataset was utilized in the subsequent evaluation of our model. It is worth mentioning that the BiLAT model is interpretable and can be used by chemists to design and synthesize compounds with improved activity. To further verify the generalization ability of the multitask BiLAT model, we also conducted another evaluation on three public datasets (Tox21, ClinTox, and SIDER). Compared with several currently popular models, BiLAT shows the best performance on two datasets. These results indicate that BiLAT is an effective tool for accelerating drug discovery.

The popularity of intelligent mobile devices has increased because they are not only convenient for people but also produce a large number of GPS trajectories. Semantic trajectories can be obtained by adding semantic information such as landmarks and activities to raw trajectories. Keyword queries in semantic trajectory databases that return the relevant places/routes have attracted increasing attention from researchers in recent years. However, existing works only consider the spatial and textual features of keywords, which cannot answer queries with temporal requirements. Simply modifying existing algorithms to support temporal requirements may lead to errors and low efficiency. Additionally, they match keywords only by string similarity without considering their semantic meanings. In this paper, we study the problem of efficient spatiotemporal keyword search in semantic trajectories (STKST). Given the position of a user and a set of keywords with temporal constraints, we aim to efficiently retrieve top-k trajectories that contain the most semantically and temporally relevant keywords and are close to the position of the user. To measure the goodness of a trajectory regarding the query, we devise a new integrated similarity measure by considering information from three aspects (spatial, temporal, and semantic). Then we develop a novel hybrid spatial–temporal–semantic index (STS-I) to organize these three kinds of information in trajectories in the form of tree structure. Finally, we propose a new algorithm STKST-I to efficiently prune unqualified trajectories based on the lower and upper bounds derived from the STS-I index. Extensive experimental studies are conducted on real trajectory datasets to verify the performance of our methods.

数据监测与控制是铝电解过程提高生产质量的重要手段,针对铝电解过程的数据监测算法缺乏多样性、实时性和稳定性等问题,研究了工艺过程数据实时聚类方法,建立了一种基于自适应生长层次神经气(GHNG)的生产奇异性监测模型.该模型包括自适应学习、节点生成与删除、拓扑结构展示等机制,为提高模型稳定性和分析数据多样性的能力,综合利用节点累积误差自适应调节获胜节点及邻域节点权值;依据在线数据演化趋势动态删除、增加神经节点并更新聚类中心位置,实现实时展现数据实例动态聚类结果,进一步提高聚类算法的时效性,同时对在线监测模型和算法进行了性能测试.最后,通过铝电解过程数据监测实例验证了该模型和算法的奇异性监测能力更强,能对铝电解工艺过程进行准确、有效的监测和控制,为生产/管理者提供决策支持.

Detecting tiny defects in cigarettes is currently a major concern for manufacturers. To address this issue, this paper investigates a hybrid model based on lightweight ViT and RCNN to provide a better balance of high performance and high accuracy. Experiments showed that the model presented in this paper has a mAP value of 85.7% at 1% of tiny defects in cigarette appearance and an inference speed of 82 FPS in an acquisition scenario with a camera resolution of 1280×280, which meets the needs of high-speed acquisition in industrial sites. The results indicate that the hybrid model can be used to detect flaws in cigarette appearance.

Nonnegative matrix factorization (NMF) methods have achieved remarkable performances in multi-view clustering due to their effectiveness and efficiency. To better obtain a low-dimensional common representation, the limited labels and the geometric structure of the multi-view data should be fully utilized in clustering. In this work, we introduce a novel multi-view learning approach, dubbed label-embedded regularized NMF with dual-graph constraints (LeNMF-DC), for clustering. Our proposed LeNMF-DC approach mainly utilizes matrix factorization to obtain a low-dimensional common representation of the multi-view data, in which the prior knowledge hidden in data can be fully explored. Specifically, we construct three graph regularization terms to preserve the manifold structure in the data, feature and label space, respectively. Moreover, we take advantage of the labels of the labeled samples without additional parameters. In addition, we develop an alternate iterative optimization scheme to solve the model of LeNMF-DC and then show its convergence rate. Compared with traditional multi-view clustering approaches, the labels of unlabeled samples in our proposed LeNMF-DC approach are assigned by the label constraint matrix rather than the clustering algorithm, and thus it avoids performance loss during the clustering. Experimental results on four benchmark datasets manifest that our LeNMF-DC approach can achieve superior performances than several state-of-the-art approaches in multi-view clustering.

This paper proposes an improved model for two-stage image defect detection in cigarette appearance that enhances both performance and accuracy. The model is based on YOLOv5s and incorporates an attention mechanism. To evaluate the model's effectiveness, we utilized a real appearance defects dataset of cigarettes. Results from the experiments demonstrate that the model can achieve a mean average precision (mAP) of 0.916 and frames per second (FPS) of 82 after 200 epoch. Additionally, in a production environment, the model demonstrated inference performance of 6.5ms (FPS 154). The high detection speed and effectiveness of the model make it suitable for on-site, real-time inspection of cigarette appearance defects detection.

The server outlet temperature is an important thermal condition to the operation of an air free-cooled data center that uses fans to continuously pass the outside air through the server room to cool the computing devices. However, the standard server's management and monitoring tool cannot read the server's built-in outlet temperature sensors fast enough to catch up the fast dynamics of the server outlet thermal condition caused by the changing server workload. Moreover, many server models do not have built-in sensors that can measure the server outlet temperature. In this paper, we develop a data-assisted first-principle model that leverages available built-in sensors and server's operating monitoring tools to achieve low-latency estimation of the server outlet temperature. Specifically, the developed model takes the inlet and processor core temperatures, server's fan speed, and processor utilization which are measured by hardware/software sensors as inputs to predict the outlet temperature with low latencies. Our extensive evaluation based on real data traces collected from a real air free-cooled data center testbed shows that our model can accurately predict the outlet temperature with an average root mean squared error ranging from 1.21 degrees C to 1.46 degrees C under various cold supply air temperatures and processor utilization levels. (C) 2021 Elsevier B.V. All rights reserved.

In analyzing dynamic characteristic of time-series data, classic prediction models rely heavily on static historical data, and tacit knowledge is difficult to be mined effectively. Therefore, a hybrid prediction model GS-GMDH is proposed based on growing neural gas (GNG) and the group method of data handling (GMDH). Firstly, a dynamic prediction mechanism, based on an incremental learning algorithm and time-series prediction, is established by GS-GMDH, by which the singularity is recognized and the prediction efficiency is improved. Secondly, to compare the performance of the proposed method, the multi-step ahead predictions with time-series data onto iron and silicon content are employed, and the new model is compared with classic machine models. Finally, the results show that the hybrid prediction model (GS-GMDH) proposed in this paper ensure an accurate and efficient prediction of time-series data for iron and silicon content.

Air free cooling is an energy-efficient cooling scheme that has been adopted in the dry and cold climate zones. To adopt this cooling scheme in Singapore's tropical condition, we designed and implemented an air free-cooled DC testbed integrating sensing and control systems for the server and room conditions. Then, we conducted extensive experiments on the testbed to understand its energy efficiency and server reliability. This paper presents the key observations, experiences, and learned lessons obtained from our testbed over a duration of nearly two years. The experiments show that (1) the air free-cooling design can achieve the power usage effectiveness of 1.05, (2) the tropics’ year-round high temperatures up to $37^\circ$ C do not impede the air free-cooling, and (3) the implementation of the air free-cooled tropical DCs requires special cares to deal with airborne contaminants to avoid fast corrosion rate and dust-induced server faults. Based on our experiment data, a set of recommendations on the temperature control and the selection of IT equipment for air free-cooled tropical DCs is made. The descriptions of the learned lessons, the resulting recommendations, and the released data can be useful to the relevant research communities, governmental agencies, and standardizing bodies.

Air free-cooled data centers (DCs) have not existed in the tropical zone due to the unique challenges of year-round high ambient temperature and relative humidity (RH). The increasing availability of servers that can tolerate higher temperatures and RH due to the regulatory bodies’ prompts to raise DC temperature setpoints sheds light upon the feasibility of air free-cooled DCs in the tropics. However, due to the complex psychrometric dynamics, operating the air free-cooled DC in the tropics generally requires adaptive control of supply air condition to maintain the computing performance and reliability of the servers. This article studies the problem of controlling the supply air temperature and RH in a free-cooled tropical DC below certain thresholds. To achieve the goal, we formulate the control problem as Markov decision processes and apply deep reinforcement learning (DRL) to learn the control policy that minimizes the cooling energy while satisfying the requirements on the supply air temperature and RH. We also develop a constrained DRL solution for performance improvements. Extensive evaluation based on real data traces collected from an air free-cooled testbed and comparisons among the unconstrained and constrained DRL approaches as well as two other baseline approaches show the superior performance of our proposed solutions.

The scheme of application (app) distribution systems involving incentivized third-party app vendors is a desirable option for the emerging edge computing systems. However, such a scheme also brings various security challenges as faced by the current mobile app distribution systems. In this paper, we study a threat named covert device association, in which the vendors of two apps collude to figure out which of their app installations run on the same edge device. If the two colluding apps are popular, the threat can be used to launch various types of further attacks at scale. For example, the user of the compromised edge device, who wishes to remain anonymous to one of the two apps, will be de-anonymized if the user is not anonymous to the other app...

The air free-cooling has been long thought infeasible in tropics due to the unique challenges of year-round high ambient temperature and relative humidity. In recent years, the increasing availability of servers that can tolerate higher temperatures and relative humidity levels sheds light upon the feasibility of the air free-cooling to enhance the data center energy efficiency. However, building an air free-cooled data center in the tropics requires extensive experiments to understand the details of how the tropical environment conditions will affect data center power consumption, computing throughput, and server hardware reliability. Thus, together with multiple partners in data center industry and research, we conducted a project that designs, builds, and experiments with an air free-cooled data center testbed consisting of three server rooms hosting 12 server racks with 60 kW total power rating. This paper presents the key observations, experiences and learned lessons obtained from our project. The experiments show that (1) the air free-cooling design that uses fans only can reduce the power usage effectiveness (PUE) by 38%, compared to the global average PUE, (2) the tropics' year-round high temperatures up to 37°C do not impede the air free-cooling, and (3) the implementation of the air free-cooled data centers in tropics requires special cares to deal with airborne contaminants to avoid fast corrosion rate and dust-induced server faults.

While thriving application (app) distribution systems involving incentivized third-party app vendors are desirable for the emerging edge computing paradigm, they also bring security challenges as faced by the current mobile app distribution systems. This article studies a threat called covert device association , in which the vendors of two apps collude to figure out which of their app installations run on the same edge device. The threat can widely spread when the two apps are popular. It is also a stepping stone for: 1) the de-anonymization attacks against the users anonymous to one of the two vendors and 2) privilege escalation in which the two colluding vendors have united privileges. We show that the threat can be implemented via a reliable and ubiquitous covert channel based on the edge device’s processor workload without requiring any privileged permissions. We present the implementation details for three attack scenarios of: 1) two Android apps; 2) an Android app and a Web session running in the mobile Tor browser; and 3) two Android Things apps. Evaluation on two smartphones and an embedded edge device shows that the covert channel gives at least 0.25 b/s data rate with zero empirical bit error rate and the covert device association can be completed within 3.2 min.

Air free-cooled data centers (DCs) have not existed in the tropical zone due to the unique challenges of year-round high ambient temperature and relative humidity (RH). The increasing availability of servers that can tolerate higher temperatures and RH due to the regulatory bodies' prompts to raise DC temperature setpoints sheds light upon the feasibility of air free-cooled DCs in tropics. This paper studies the problem of controlling the temperature and RH of the air supplied to the servers in a free-cooled tropical DC below certain thresholds to maintain servers' computing performance and reliability. To achieve the goal, a portion of the hot air generated by the servers is recirculated and mixed with the fresh outside air to adjust the RH of the supply air. To address the complex psychrometric dynamics, we apply deep reinforcement learning to learn the control policy that aims at minimizing the energy used for moving air and on-demand cooling. Extensive evaluation based on real data traces collected from an air free-cooled testbed and comparisons with hysteresis-based and model-predictive control approaches show the superior performance of our solution.