Deep approaches to anomaly detection have recently shown promising results over shallow methods on large and complex datasets. Typically is treated as an unsupervised learning problem. In practice however, one may have---in addition a set of unlabeled samples---access small pool labeled samples, e.g. subset verified by some domain expert being normal or anomalous. Semi-supervised aim utilize such but most proposed are limited merely including samples. Only few take advantage anomalies, with...

Abstract Deep learning has revolutionized data science in many fields by greatly improving prediction performances comparison to conventional approaches. Recently, explainable artificial intelligence emerged as an area of research that goes beyond pure improvement extracting knowledge from deep methodologies through the interpretation their results. We investigate such explanations explore genetic architectures phenotypes genome-wide association studies. Instead testing each position genome...

Anomaly detection for network intrusion is usually considered an unsupervised task. Prominent techniques, such as one-class support vector machines, learn a hypersphere enclosing data, mapped to space, that points outside of the ball are anomalous. However, this setup ignores relevant information expert and background knowledge. In paper, we rephrase anomaly active learning We propose effective strategy query low-confidence observations expand data basis with minimal labeling effort. Our...

Abstract In many research areas scientists are interested in clustering objects within small datasets while making use of prior knowledge from large reference datasets. We propose a method to apply the machine learning concept transfer unsupervised problems and show its effectiveness field single-cell RNA sequencing (scRNA-Seq). The goal scRNA-Seq experiments is often definition cataloguing cell types transcriptional output individual cells. To improve disease- or tissue-specific datasets,...

We present ClusterSVDD, a methodology that unifies support vector data descriptions (SVDDs) and k-means clustering into single formulation. This allows both methods to benefit from one another, i.e., by adding flexibility using multiple spheres for SVDDs increasing anomaly resistance through kernels k-means. In particular, our approach leads new interpretation of as regularized mode seeking algorithm. The unifying formulation further deriving algorithms transferring knowledge oneclass...

In order to solve large-scale lasso problems, screening algorithms have been developed that discard features with zero coefficients based on a computationally efficient rule. Most existing rules were from spherical constraint and half-space constraints dual optimal solution. However, admit at most two due the computational cost incurred by half-spaces, even though additional may be useful more features. this paper, we present AdaScreen, an adaptive rule ensemble, which allows combine any one...

The task of structured output prediction deals with learning general functional dependencies between arbitrary input and spaces. In this context, two loss-sensitive formulations for maximum-margin training have been proposed in the literature, which are referred to as margin slack rescaling, respectively. latter is believed be more accurate easier handle. Nevertheless, it not popular due lack known efficient inference algorithms; therefore, rescaling--which requires a similar type normal...

Abstract We present Oqtans, an open-source workbench for quantitative transcriptome analysis, that is integrated in Galaxy. Its distinguishing features include customizable computational workflows and a modular pipeline architecture facilitates comparative assessment of tool data quality. Oqtans integrates assortment machine learning-powered tools into Galaxy, which show superior or equal performance to state-of-the-art tools. Implemented comprise complete analysis workflow: short-read...

Authorship detection is a challenging task due to many design choices the user has decide on. The performance highly depends on right set of features, amount data, in-sample vs. out-of-sample settings, and profile- instance-based approaches. So far, variety combinations renders off-the-shelf methods for authorship inappropriate. We propose novel generally deployable method that does not share these limitations. treat attribution as an anomaly problem where author regions are learned in...

Complex problems may require sophisticated, non-linear learning methods such as kernel machines or deep neural networks to achieve state of the art prediction accuracies. However, high accuracies are not only objective consider when solving using machine learning. Instead, particular scientific applications some explanation learned function. Unfortunately, most do come with out box straight forward interpretation. Even linear functions explain if features exhibit complex correlation...

Identifying discriminative motifs underlying the functionality and evolution of organisms is a major challenge in computational biology. Machine learning approaches such as support vector machines (SVMs) achieve state-of-the-art performances genomic discrimination tasks, but—due to its black-box character—motifs decision function are largely unknown. As remedy, positional oligomer importance matrices (POIMs) allow us visualize significance position-specific subsequences. Although being step...

High prediction accuracies are not the only objective to consider when solving problems using machine learning. Instead, particular scientific applications require some explanation of learned function. For computational biology, positional oligomer importance matrices (POIMs) have been successfully applied explain decision support vector machines (SVMs) weighted-degree (WD) kernels. To extract relevant biological motifs from POIMs, motifPOIM method has devised and showed promising results on...

Analyzing data with latent spatial and/or temporal structure is a challenge for machine learning. In this paper, we propose novel nonlinear model studying dependence structure. It successfully combines the concepts of Markov random fields, transductive learning, and regression, making heavy use notion joint feature maps. Our conditional field regression able to infer states by combining limited labeled high precision unlabeled containing measurement uncertainty. manner, can propagate...

Conventionally, neuroscientific data is analyzed based on the behavioral response of participant. This approach assumes that errors participants are in line with neural processing. However, this may not be case, particular experiments time pressure or studies investigating threshold perception. In these cases, error distribution deviates from uniformity due to heteroscedastic nature underlying experimental set-up. problem systematic and structured (non-uniform) label noise ignored when...

The last years have seen a rise of interest in using electroencephalography-based brain computer interfacing methodology for investigating non-medical questions, beyond the purpose communication and control. One these novel applications is to examine how signal quality being processed neurally, which particular industry, besides providing neuroscientific insights. As most behavioral experiments neurosciences, assessment given stimulus by subject required. Based on an EEG study speech...