Explore the vast range of titles available.

Time Series Classification (TSC) is an important and challenging problem in data mining. With the increase of time series data availability, hundreds of TSC algorithms have been proposed. Among these methods, only a few have considered Deep Neural Networks (DNNs) to perform this task. This is surprising as deep learning has seen very successful applications in the last years. DNNs have indeed revolutionized the field of computer vision especially with the advent of novel deeper architectures such as Residual and Convolutional Neural Networks. Apart from images, sequential data such as text and audio can also be processed with DNNs to reach state-of-the-art performance for document classification and speech recognition. In this article, we study the current state-of-the-art performance of deep learning algorithms for TSC by presenting an empirical study of the most recent DNN architectures for TSC. We give an overview of the most successful deep learning applications in various time series domains under a unified taxonomy of DNNs for TSC. We also provide an open source deep learning framework to the TSC community where we implemented each of the compared approaches and evaluated them on a univariate TSC benchmark (the UCR/UEA archive) and 12 multivariate time series datasets. By training 8730 deep learning models on 97 time series datasets, we propose the most exhaustive study of DNNs for TSC to date.

Time series are ubiquitous in data mining applications. Similar to other types of data, annotations can be challenging to acquire, thus preventing from training time series classification models. In this context, clustering methods can be an appropriate alternative as they create homogeneous groups allowing a better analysis of the data structure. Time series clustering has been investigated for many years and multiple approaches have already been proposed. Following the advent of deep learning in computer vision, researchers recently started to study the use of deep clustering to cluster time series data. The existing approaches mostly rely on representation learning (imported from computer vision), which consists of learning a representation of the data and performing the clustering task using this new representation. The goal of this paper is to provide a careful study and an experimental comparison of the existing literature on time series representation learning for deep clustering. In this paper, we went beyond the sole comparison of existing approaches and proposed to decompose deep clustering methods into three main components: (1) network architecture, (2) pretext loss, and (3) clustering loss. We evaluated all combinations of these components (totaling 300 different models) with the objective to study their relative influence on the clustering performance. We also experimentally compared the most efficient combinations we identified with existing non-deep clustering methods. Experiments were performed using the largest repository of time series datasets (the UCR/UEA archive) composed of 128 univariate and 30 multivariate datasets. Finally, we proposed an extension of the class activation maps method to the unsupervised case which allows to identify patterns providing highlights on how the network clustered the time series.

Dynamic Time Warping (DTW) is a highly competitive distance measure for most time series data mining problems. Obtaining the best performance from DTW requires setting its only parameter, the maximum amount of warping (w). In the supervised case with ample data, w is typically set by cross-validation in the training stage. However, this method is likely to yield suboptimal results for small training sets. For the unsupervised case, learning via cross-validation is not possible because we do not have access to labeled data. Many practitioners have thus resorted to assuming that “the larger the better”, and they use the largest value of w permitted by the computational resources. However, as we will show, in most circumstances, this is a naïve approach that produces inferior clusterings. Moreover, the best warping window width is generally non-transferable between the two tasks, i.e., for a single dataset, practitioners cannot simply apply the best w learned for classification on clustering or vice versa. In addition, we will demonstrate that the appropriate amount of warping not only depends on the data structure, but also on the dataset size. Thus, even if a practitioner knows the best setting for a given dataset, they will likely be at a lost if they apply that setting on a bigger size version of that data. All these issues seem largely unknown or at least unappreciated in the community. In this work, we demonstrate the importance of setting DTW’s warping window width correctly, and we also propose novel methods to learn this parameter in both supervised and unsupervised settings. The algorithms we propose to learn w can produce significant improvements in classification accuracy and clustering quality. We demonstrate the correctness of our novel observations and the utility of our ideas by testing them with more than one hundred publicly available datasets. Our forceful results allow us to make a perhaps unexpected claim; an underappreciated “low hanging fruit” in optimizing DTW’s performance can produce improvements that make it an even stronger baseline, closing most or all the improvement gap of the more sophisticated methods proposed in recent years.

In the context of global change, up-to-date land use land cover (LULC) maps is a major challenge to assess pressures on natural areas. These maps also allow us to assess the evolution of land cover and to quantify changes over time (such as urban sprawl), which is essential for having a precise understanding of a given territory. Few studies have combined information from Sentinel-1 and Sentinel-2 imagery, but merging radar and optical imagery has been shown to have several benefits for a range of study cases, such as semantic segmentation or classification. For this study, we used a newly produced dataset, MultiSenGE, which provides a set of multitemporal and multimodal patches over the Grand-Est region in France. To merge these data, we propose a CNN approach based on spatio-temporal and spatio-spectral feature fusion, ConvLSTM+Inception-S1S2. We used a U-Net base model and ConvLSTM extractor for spatio-temporal features and an inception module for the spatio-spectral features extractor. The results show that describing an overrepresented class is preferable to map urban fabrics (UF). Furthermore, the addition of an Inception module on a date allowing the extraction of spatio-spectral features improves the classification results. Spatio-spectro-temporal method (ConvLSTM+Inception-S1S2) achieves higher global weighted F1Score than all other methods tested.

Human intelligence is evident in the concept of clinical reasoning, which has been defined as \"the internal mental processes that a physician uses when approaching clinical situations.\" This central component of physicians'; competence, once honed, allows them to make diagnoses. In medicine, clinical reasoning is often understood from the perspective of cognitive psychology's information process theory. Artificial intelligence (AI) may refer to several different methods. Most AI diagnostics are based on machine learning algorithms that are \"intelligent\" enough to handle difficult and complex problems; algorithms rely on human intelligence for their creation. Recently, substantial progress has been made in this field through the resurgence of neural networks--a family of methods of machine learning--and particularly deep neural networks. Here, Pelaccia et al focus mainly on machine learning (specifically deep neural networks). They analyze the differences in the ways humans and AI approach diagnostic reasoning to argue that human reasoning will not become obsolete in medical diagnosis.

Our aim is to complement observer-dependent approaches of immune cell evaluation in microscopy images with reproducible measures for spatial composition of lymphocytic infiltrates. Analyzing such patterns of inflammation is becoming increasingly important for therapeutic decisions, for example in transplantation medicine or cancer immunology. We developed a graph-based assessment of lymphocyte clustering in full whole slide images. Based on cell coordinates detected in the full image, a Delaunay triangulation and distance criteria are used to build neighborhood graphs. The composition of nodes and edges are used for classification, e.g. using a support vector machine. We describe the variability of these infiltrates on CD3/CD20 duplex staining in renal biopsies of long-term functioning allografts, in breast cancer cases, and in lung tissue of cystic fibrosis patients. The assessment includes automated cell detection, identification of regions of interest, and classification of lymphocytic clusters according to their degree of organization. We propose a neighborhood feature which considers the occurrence of edges with a certain type in the graph to distinguish between phenotypically different immune infiltrates. Our work addresses a medical need and provides a scalable framework that can be easily adjusted to the requirements of different research questions.

The development of very high-resolution (VHR) satellite imagery has produced a huge amount of data. The multiplication of satellites which embed different types of sensors provides a lot of heterogeneous images. Consequently, the image analyst has often many different images available, representing the same area of the Earth surface. These images can be from different dates, produced by different sensors, or even at different resolutions. The lack of machine learning tools using all these representations in an overall process constraints to a sequential analysis of these various images. In order to use all the information available simultaneously, we propose a framework where different algorithms can use different views of the scene. Each one works on a different remotely sensed image and, thus, produces different and useful information. These algorithms work together in a collaborative way through an automatic and mutual refinement of their results, so that all the results have almost the same number of clusters, which are statistically similar. Finally, a unique result is produced, representing a consensus among the information obtained by each clustering method on its own image. The unified result and the complementarity of the single results (i.e., the agreement between the clustering methods as well as the disagreement) lead to a better understanding of the scene. The experiments carried out on multispectral remote sensing images have shown that this method is efficient to extract relevant information and to improve the scene understanding.

Interventional healthcare will evolve from an artisanal craft based on the individual experiences, preferences and traditions of physicians into a discipline that relies on objective decision-making on the basis of large-scale data from heterogeneous sources.

Time series averages are one key input to temporal data mining techniques such as classification, clustering, forecasting, etc. In practice, the optimality of estimated averages often impacts the performance of such temporal data mining techniques. Practically, an estimated average is presumed to be optimal if it minimizes the discrepancy between itself and members of an averaged set while preserving descriptive shapes. However, estimating an average under such constraints is often not trivial due to temporal shifts. To this end, all pioneering averaging techniques propose to align averaged series before estimating an average. Practically, the alignment gets performed to transform the averaged series, such that, after the transformation, they get registered to their arithmetic mean. However, in practice, most proposed alignment techniques often introduce additional challenges. For instance, Dynamic Time Warping (DTW)-based alignment techniques make the average estimation process non-smooth, non-convex, and computationally demanding. With such observation in mind, we approach time series averaging as a generative problem. Thus, we propose to mimic the effects of temporal alignment in the latent space of multi-tasking neural networks. We also propose to estimate (augment) time domain averages from the latent space representations. With this approach, we provide state-of-the-art latent space registration. Moreover, we provide time domain estimations that are better than the estimates generated by some pioneering averaging techniques.