Data Mining Practical Machine Learning Tools and Techniques

The opportunities for bacterial population genomics that are being realised by the application of parallel nucleotide sequencing require novel bioinformatics platforms These must be capable of the storage, retrieval, and analysis of linked phenotypic and genotypic information in an accessible, scalable and computationally efficient manner The Bacterial Isolate Genome Sequence Database (BIGSDB) is a scalable, open source, web-accessible database system that meets these needs, enabling phenotype and sequence data, which can range from a single sequence read to whole genome data, to be efficiently linked for a limitless number of bacterial specimens The system builds on the widely used mlstdbNet software, developed for the storage and distribution of multilocus sequence typing (MLST) data, and incorporates the capacity to define and identify any number of loci and genetic variants at those loci within the stored nucleotide sequences These loci can be further organised into 'schemes' for isolate characterisation or for evolutionary or functional analyses Isolates and loci can be indexed by multiple names and any number of alternative schemes can be accommodated, enabling cross-referencing of different studies and approaches LIMS functionality of the software enables linkage to and organisation of laboratory samples The data are easily linked to external databases and fine-grained authentication of access permits multiple users to participate in community annotation by setting up or contributing to different schemes within the database Some of the applications of BIGSDB are illustrated with the genera Neisseria and Streptococcus The BIGSDB source code and documentation are available at http://pubmlstorg/software/database/bigsdb/
  Genomic data can be used to characterise bacterial isolates in many different ways but it can also be efficiently exploited for evolutionary or functional studies BIGSDB represents a freely available resource that will assist the broader community in the elucidation of the structure and function of bacteria by means of a population genomics approach

/pdf/bigsdb-scalable-analysis-of-bacterial-genome-variation-at-4zefepabii.pdf

BIGSdb: Scalable analysis of bacterial genome variation at the population level

We introduce TUT Acoustic Scenes 2016 database for environmental sound research, consisting of binaural recordings from 15 different acoustic environments. A subset of this database, called TUT Sound Events 2016, contains annotations for individual sound events, specifically created for sound event detection. TUT Sound Events 2016 consists of residential area and home environments, and is manually annotated to mark onset, offset and label of sound events. In this paper we present the recording and annotation procedure, the database content, a recommended cross-validation setup and performance of supervised acoustic scene classification system and event detection baseline system using mel frequency cepstral coefficients and Gaussian mixture models. The database is publicly released to provide support for algorithm development and common ground for comparison of different techniques.

TUT database for acoustic scene classification and sound event detection

This paper presents and discusses various metrics proposed for evaluation of polyphonic sound event detection systems used in realistic situations where there are typically multiple sound sources active simultaneously The system output in this case contains overlapping events, marked as multiple sounds detected as being active at the same time The polyphonic system output requires a suitable procedure for evaluation against a reference Metrics from neighboring fields such as speech recognition and speaker diarization can be used, but they need to be partially redefined to deal with the overlapping events We present a review of the most common metrics in the field and the way they are adapted and interpreted in the polyphonic case We discuss segment-based and event-based definitions of each metric and explain the consequences of instance-based and class-based averaging using a case study In parallel, we provide a toolbox containing implementations of presented metrics

Metrics for Polyphonic Sound Event Detection

With the decrease of DNA sequencing costs, sequence-based typing methods are rapidly becoming the gold standard for epidemiological surveillance. These methods provide reproducible and comparable results needed for a global scale bacterial population analysis, while retaining their usefulness for local epidemiological surveys. Online databases that collect the generated allelic profiles and associated epidemiological data are available but this wealth of data remains underused and are frequently poorly annotated since no user-friendly tool exists to analyze and explore it. PHYLOViZ is platform independent Java software that allows the integrated analysis of sequence-based typing methods, including SNP data generated from whole genome sequence approaches, and associated epidemiological data. goeBURST and its Minimum Spanning Tree expansion are used for visualizing the possible evolutionary relationships between isolates. The results can be displayed as an annotated graph overlaying the query results of any other epidemiological data available. PHYLOViZ is a user-friendly software that allows the combined analysis of multiple data sources for microbial epidemiological and population studies. It is freely available at http://www.phyloviz.net
 .

/pdf/phyloviz-phylogenetic-inference-and-data-visualization-for-5blz210ltf.pdf

PHYLOViZ: phylogenetic inference and data visualization for sequence based typing methods

Background: Multilocus Sequence Typing (MLST) is a frequently used typing method for the analysis of the clonal relationships among strains of several clinically relevant microbial species. MLST is based on the sequence of housekeeping genes that result in each strain having a distinct numerical allelic profile, which is abbreviated to a unique identifier: the sequence type (ST). The relatedness between two strains can then be inferred by the differences between allelic profiles. For a more comprehensive analysis of the possible patterns of evolutionary descent, a set of rules were proposed and implemented in the eBURST algorithm. These rules allow the division of a data set into several clusters of related strains, dubbed clonal complexes, by implementing a simple model of clonal expansion and diversification. Within each clonal complex, the rules identify which links between STs correspond to the most probable pattern of descent. However, the eBURST algorithm is not globally optimized, which can result in links, within the clonal complexes, that violate the rules proposed. Results: Here, we present a globally optimized implementation of the eBURST algorithm – goeBURST. The search for a global optimal solution led to the formalization of the problem as a graphic matroid, for which greedy algorithms that provide an optimal solution exist. Several public data sets of MLST data were tested and differences between the two implementations were found and are discussed for five bacterial species: Enterococcus faecium, Streptococcus pneumoniae, Burkholderia pseudomallei, Campylobacter jejuni and Neisseria spp.. A novel feature implemented in goeBURST is the representation of the level of tiebreak rule reached before deciding if a link should be drawn, which can used to visually evaluate the reliability of the represented hypothetical pattern of descent. Conclusion: goeBURST is a globally optimized implementation of the eBURST algorithm, that identifies alternative patterns of descent for several bacterial species. Furthermore, the algorithm can be applied to any multilocus typing data based on the number of differences between numeric profiles. A software implementation is available at http://goeBURST.phyloviz.net.

/pdf/global-optimal-eburst-analysis-of-multilocus-typing-data-5fxit7pghy.pdf

Global optimal eBURST analysis of multilocus typing data using a graphic matroid approach

In this paper, a novel approach to video temporal decomposition into semantic units, termed scenes, is presented. In contrast to previous temporal segmentation approaches that employ mostly low-level visual or audiovisual features, we introduce a technique that jointly exploits low-level and high-level features automatically extracted from the visual and the auditory channel. This technique is built upon the well-known method of the scene transition graph (STG), first by introducing a new STG approximation that features reduced computational cost, and then by extending the unimodal STG-based temporal segmentation technique to a method for multimodal scene segmentation. The latter exploits, among others, the results of a large number of TRECVID-type trained visual concept detectors and audio event detectors, and is based on a probabilistic merging process that combines multiple individual STGs while at the same time diminishing the need for selecting and fine-tuning several STG construction parameters. The proposed approach is evaluated on three test datasets, comprising TRECVID documentary films, movies, and news-related videos, respectively. The experimental results demonstrate the improved performance of the proposed approach in comparison to other unimodal and multimodal techniques of the relevant literature and highlight the contribution of high-level audiovisual features toward improved video segmentation to scenes.

https://mklab.iti.gr/files/csvt11_preprint.pdf

Temporal Video Segmentation to Scenes Using High-Level Audiovisual Features

Audio event detection is one of the tasks of the European project VIDIVIDEO. This paper focuses on the detection of non-speech events, and as such only searches for events in audio segments that have been previously classified as non-speech. Preliminary experiments with a small corpus of sound effects have shown the potential of this type of corpus for training purposes. This paper describes our experiments with SVM and HMM-based classifiers, using a 290-hour corpus of sound effects. Although we have only built detectors for 15 semantic concepts so far, the method seems easily portable to other concepts. The paper reports experiments with multiple features, different kernels and several analysis windows. Preliminary experiments on documentaries and films yielded promising results, despite the difficulties posed by the mixtures of audio events that characterize real sounds.

Non-speech audio event detection

In this paper we describe our TRECVID 2009 video re- trieval experiments. The MediaMill team participated in three tasks: concept detection, automatic search, and in- teractive search. The starting point for the MediaMill con- cept detection approach is our top-performing bag-of-words system of last year, which uses multiple color descriptors, codebooks with soft-assignment, and kernel-based supervised learning. We improve upon this baseline system by explor- ing two novel research directions. Firstly, we study a multi- modal extension by including 20 audio concepts and fusion using two novel multi-kernel supervised learning methods. Secondly, with the help of recently proposed algorithmic re- nements of bag-of-word representations, a GPU implemen- tation, and compute clusters, we scale-up the amount of vi- sual information analyzed by an order of magnitude, to a total of 1,000,000 i-frames. Our experiments evaluate the merit of these new components, ultimately leading to 64 ro- bust concept detectors for video retrieval. For retrieval, a robust but limited set of concept detectors justi es the need to rely on as many auxiliary information channels as pos- sible. For automatic search we therefore explore how we can learn to rank various information channels simultane- ously to maximize video search results for a given topic. To further improve the video retrieval results, our interactive search experiments investigate the roles of visualizing pre- view results for a certain browse-dimension and relevance feedback mechanisms that learn to solve complex search top- ics by analysis from user browsing behavior. The 2009 edi- tion of the TRECVID benchmark has again been a fruitful participation for the MediaMill team, resulting in the top ranking for both concept detection and interactive search. Again a lot has been learned during this year's TRECVID campaign; we highlight the most important lessons at the end of this paper.

/pdf/the-mediamill-trecvid-2009-semantic-video-search-engine-3c8y07oh4j.pdf

The MediaMill TRECVID 2009 Semantic Video Search Engine

This paper describes our work on audio event detection, one of our tasks in the European project VIDIVIDEO. Preliminary experiments with a small corpus of sound effects have shown the potential of this type of corpus for training purposes. This paper describes our experiments with SVM classifiers, and different features, using a 290-hour corpus of sound effects, which allowed us to build detectors for almost 50 semantic concepts. Although the performance of these detectors on the development set is quite good (achieving an average F-measure of 0.87), preliminary experiments on documentaries and films showed that the task is much harder in real-life videos, which so often include overlapping audio events. Index Terms: event detection, audio segmentation

http://doutoramento.schiu.com/referencias/outras/Bugalho,%20M.;%20Portelo,%20J.;%20Trancoso,%20I.;%20Pellegrini,%20T.;%20Abad,%20A.%20-%20Detecting%20Audio.pdf

Miguel Bugalho

Papers

Global optimal eBURST analysis of multilocus typing data using a graphic matroid approach

Temporal Video Segmentation to Scenes Using High-Level Audiovisual Features

Non-speech audio event detection

The MediaMill TRECVID 2009 Semantic Video Search Engine

Detecting Audio Events for Semantic Video Search