Shapelets are interclass discriminative subsequences that can be used to characterize target classes. Learning shapelets by continuous optimization has recently been studied to improve classification accuracy. However, there are two issues in previous studies. First, since the locations where shapelets appear in the time series are determined by only their shapes, shapelets may appear at incorrect and non-discriminative locations in the time series, degrading the accuracy and interpretability. Second, the theoretical interpretation of learned shapelets has been limited to binary classification. To tackle the first issue, we propose a continuous optimization that learns not only shapelets but also their probable locations in a time series, and we show theoretically that this enhances feature discriminability. To tackle the second issue, we provide a theoretical interpretation of shapelet closeness to the time series for target / off-target classes when learning with softmax loss, which allows for multi-class classification. We demonstrate the effectiveness of the proposed method in terms of accuracy, runtime, and interpretability on the UCR archive.

The need for explainability in time-series classification models has been increasing. Counterfactual explanations recommend how to modify the features of an original instance so that the prediction by a given classifier flips to the desired class. Since features in the time series are temporally dependent, interpretability is improved by considering intervals where the counterfactual can deviate from the original instance. In this study, we propose a model-agnostic counterfactual generation method (CEI) that jointly learns these intervals and the counterfactual. Furthermore, CEI can generate a counterfactual tailored to the directly specified limited number of intervals. We mathematically formulate CEI as a continuous optimization and demonstrate its effectiveness on the UCR datasets.

Shapelets are subsequences that are effective for classifying time-series instances. Learning shapelets by a continuous optimization has recently been studied to improve computational efficiency and classification performance. However, existing methods have employed predefined and fixed shapelet lengths during the continuous optimization, despite the fact that shapelets and their lengths are inherently interdependent and thus should be jointly optimized. To efficiently explore shapelets of high quality in terms of interpretability and inter-class separability, this study makes the shapelet lengths continuous and learnable. The proposed formulation jointly optimizes not only a binary classifier and shapelets but also shapelet lengths. The derived SGD optimization can be theoretically interpreted as improving the quality of shapelets in terms of shapelet closeness to the time series for target / off-target classes. We demonstrate improvements in area under the curve, total training time, and shapelet interpretability on UCR binary datasets.

Shapelets are subsequences that are effective for classifying time-series instances. In this study, we consider when each time-series instance is obtained as progress, and formulate the problem of learning shapelet evolution over progress. For example, shapelets can change their shapes according to progress with human habituation, seasonal effects, and system degradation. When given time-series instances, progress values, and binary class labels, the proposed optimization formulation can jointly learn not only the shapelets and a classifier but also regression models for predicting shapelet evolution. The derived optimization solution method allows regression models to be learned by using off-the-shelf regression solvers, and scales linearly with time-series length. We demonstrate its effectiveness in industrial case studies.

Shapelets are subsequences that are effective for classifying time-series instances. Joint learning of both classifiers and shapelets has recently been studied because this approach improves algorithmic complexity and classification performance. However, the existing methods lack the power of feature discrimination due to using traditional sigmoid cross-entropy loss functions. To enhance feature discriminability, we propose self-adaptive scaling of the loss functions, inspired by the recent discriminative loss in computer vision. In addition, we propose a theoretically sound regularization that enhances feature discriminability and maintains shapelet interpretability by shrinking appropriate features. Using UCR datasets, we demonstrate improved area under the curve and interpretability of shapelets with a small number of shapelets.

Shapelets are time-series segments effective for classifying time-series instances.Joint learning of both classifiers and shapelets has been studied in recent years because this approach provides both superior classification performance and interpretable results.However, the optimization formulation is nonconvex, so bad local minima must be avoided.Very recently, this issue has been tackled by introducing Self-Paced Learning (SPL) into these methods.With the aim of intelligently discovering initial shapelets, we introduce two steps into this binary classification method so as to consistently optimize the same loss function of interest: optimizing discovery of discriminative initial shapelets from many time-series segments by using supervised feature selection, and jointly optimizing shapelets, model parameters, and latent instance weights in SPL.Using UCR datasets, we demonstrate improved area under the curve, and interpretability of shapelets where the number of shapelets is small.

Large amounts of time-series data have become accessible due to the rapid development of Internet-of-Things technologies and the demand for extracting useful knowledge from these data is increasing. Toward this goal, time-series segmentation — dividing data into similar segments — is a promising method for understanding the mechanisms in time-series data. In this paper, we focus on time-lag that appears in real datasets. Time lag — a typical phenomenon in time-series data — occurs when the speed of information diffusion differs between variables. However, conventional methods cannot distinguish differences in segmentation positions. In response, we propose Lag-Aware Multivariate Time-Series Segmentation (LAMTSS), an algorithm capturing time-lag across variables to determine segmentation positions for each variable. LAMTSS utilizes dynamic time warping without hyperparameter tuning. We confirm the accuracy of LAMTSS using artificial datasets and demonstrate the discovery of useful knowledge in real datasets.

Shapelets are time-series segments effective for classifying time-series instances. Joint learning of both classifiers and shapelets has been studied in recent years, because such methods provide both interpretable results and superior accuracy. Partial Area Under the ROC curve (pAUC) for a low range of False Positive Rates (FPR) is an important performance measure for practical cases in industries such as medicine, manufacturing, and maintenance. In this study, we propose a method that jointly learns both shapelets and a classifier for pAUC optimization in any FPR range, including the full AUC. We demonstrate superiority of pAUC on UCR time-series datasets and its effectiveness in industrial case studies.

There is a need for forecasting of short-range future values in data streams such as traffic flows, stock prices, and electricity consumption. However, concept drift in non-stationary data streams is an important problem. We propose an online prediction method called OPOSSAM for such data streams. OPOSSAM manages time-series segments in short-term memory and long-term memory, and forecasts future values by local regression based on the similarity of time-series segments. In particular, OPOSSAM keeps long-term memory consistent by reducing redundant samples with large prediction errors, and automatically adjusts the prediction model based on short-term memory from the prior model learned from the entire memory in order to deal with concept drift. Experimental results show accuracy superior to that of baseline methods on real-world datasets of traffic flow, stock prices, and electricity consumption.

Shapelets are time-series segments effective for classifying time-series datasets. In recent years, the discovery of shapelets by classifier learning has been studied. Methods for shapelet discovery have attracted great interest because they provide not only interpretable results but also superior classifi-cation performance. However, they do not consider imbalanced classifications between majority and minority classes, which may occur in actual applications (e.g., anomaly detection). Our aim is to learn shapelets and classifiers using only training data for the majority class without the minority class. We propose a method called one-class learning time-series shapelets (OCLTS). OCLTS efficiently and simultaneously optimizes both the shapelets and a non-linear classifier based on a one-class support vector machine by a stochastic sub-gradient descent algorithm. Experimental results show the method's effectiveness for interpretability and imbalanced binary classification.

This paper proposes the improvement method of operability in a bilateral control by using visual information. The bilateral control system consists of a master system which is manipulated by the operator and a slave system which contacts remove environment. Considering the actual use of the bilateral control, the operator manipulates the master system while watching the visual information from the slave side. In this situation, there is the time delay of visual information. This time delay has the bad influence of the operability. In order to improve the operability, the synchronization method of visual and tactile information is described. The virtual model, which consists of the virtual slave model and the virtual object model, is proposed to compensate this time delay. This virtual model is overlaid on the visual image from the slave side. As a result, the operator grasps the motion of the slave system and the object without time delay. The validity of the proposed system was confirmed from the experimental results.

This paper presents a new tourism local community system based on the LOD (Linked Open Data). In this study, the tourism-related data are converted into the RDF (Resource Description Framework) data and stored as LOD database in the local community system. N-triple data are generated by parsing the texts and connected to other related terms. Therefore, the relationship could be realized close to the ontology. The prototype system is constructed and the validity of the proposed system is confirmed by the experiments.

This paper proposes a synchronization method of visual and tactile information by using a virtual slave model in a bilateral control. Considering an actual remote-operated system, an operator operates a master system while watching visual information that is sent from a slave side. However, transmission of position, force, and visual information has communication delay. This is because the slave system on the image moves later than the actual slave system. This communication delay gives a bad effect on operability in the bilateral control. Therefore, the synchronization method of visual and tactile information is important. From this viewpoint, visual and tactile information in bilateral control is synchronized by using the virtual slave model. This virtual slave model is overlaid on the image from the slave side. The operator operates the master system while watching the virtual slave model overlaid on the image. As a result, the operability is improved, since the virtual slave model helps the communication delay of visual information. The effectiveness of the proposed system was confirmed form the experimental results.

Data stream management systems (DSMSs) are useful for the management and processing of continuous data at a high input rate with low latency. In the automotive domain, embedded systems use a variety of sensor data and communications from outside the vehicle to promote autonomous and safe driving. Thus, the software developed for these systems must be capable of handling large volumes of data and complex processing. At present, we are developing a platform for the integration and management of data in an automotive embedded system using a DSMS. However, compared with conventional DSMS fields, we have encountered new challenges such as precompiling queries when designing automotive systems (which demands time predictability), distributed stream processing in in-vehicle networks, and real-time scheduling and sensor data fusion by stream processing. Therefore, we developed an automotive embedded DSMS (AEDSMS) to address these challenges. The main contributions of the present study are: (1) a clear understanding of the challenges faced when introducing DSMSs into the automotive field; (2) the development of AEDSMS to tackle these challenges; and (3) an evaluation of AEDSMS during runtime using a driving assistance application.

Automotive applications are typical cyber-physical systems, which perform real-time continuous data processing using a variety of onboard sensors and communications from outside the vehicle. However, outside-the-vehicle data transmissions often introduce significant data rate fluctuations, where arrival times can vary or may not be guaranteed. In this study, we investigate real-time data stream processing for automotive applications based on earliest deadline first (EDF) scheduling. When sensor data with an early deadline arrive late to a data stream management system (DSMS), the EDF scheduler enqueues the late data as if they had arrived earlier. As a result, data streams are preemptable, and the stream queues do not satisfy FIFO because they are out-of-order. However, existing real-time scheduling of data streams cannot handle out-of-order queues, and searching of the out-of-order queues based on EDF degrades the performance owing to frequent accessing of the queues. In this study, we present efficient EDF scheduling for the out-of-order stream queues (i.e., Preemptable data streams) in the DSMS. The main contributions of this study are: (1) a seamless definition of EDF scheduling for preemptable data streams (EDF-PStream), which is based on the definition of general data stream processing, (2) a proposal of a reasonable task design for EDF-PStream by merging operators, and (3) a runtime evaluation of EDF-PStream using automotive applications, this includes a comparison with data stream scheduling methods.

Recently, there have been strong demand and interest for developing methods to analyze driving data for extracting traffic safety information. In automobile research field, several methods for detecting sudden braking have been proposed; however, these methods cannot answer the question what is the causes of such sudden braking events. In previous research, we have proposed a method to estimate incidents locations which interfered with smooth driving by using smartphone, but the method just works for the cases of vehicle stop. In this paper, we propose a new braking model which can estimate incidents locations caused sudden braking for both cases of vehicle stop and non-stop. We take real world experiments in order to validate the incidents map result. The result shows that based on the proposed method, incidents map is accurately achieved.

An electro-conjugate fluid (ECF) is a kind of functional fluid, which produces a jet flow (ECF jet) when subjected to high DC voltage. It is known that a strong ECF jet is generated under nonuniform electric field, for example, the field with a pair of needle and ring electrodes. This study introduces the ECF jet to develop a novel flexible robot hand. First, we characterize the ECF jet generator which could be a micro fluid pressure source of the robot hand, and confirm the effect of the variation of electrode parameters and the number of electrode pairs on its performance. Next, we investigate the characteristics of the robot finger which mainly consists of the ECF jet generator, a flexible rubber finger and an ECF tank. The robot finger is integrated with the pressure source (ECF jet generator) and the tank, and is successfully driven. Finally we developed a five-fingered flexible robot hand and demonstrate that the robot hand can grasp some objects with various shapes without any complex controller. The height, the width and the mass of the robot hand are approximately 60 mm, 40 mm and 15 g, respectively.