We consider the problem of designing models to leverage a recently introduced approximate model averaging technique called dropout. We define a simple new model called maxout (so named because its output is the max of a set of inputs, and because it is a natural companion to dropout) designed to both facilitate optimization by dropout and improve the accuracy of dropout's fast approximate model averaging technique. We empirically verify that the model successfully accomplishes both of these tasks. We use maxout and dropout to demonstrate state of the art classification performance on four benchmark datasets: MNIST, CIFAR-10, CIFAR-100, and SVHN.

Linear Discriminant Analysis (LDA) is a well-known method for dimensionality reduction and classification. Previous studies have also extended the binary-class case into multi-classes. However, many applications, such as object detection and keyframe extraction cannot provide consistent instance-label pairs, while LDA requires labels on instance level for training. Thus it cannot be directly applied for semi-supervised classification problem. In this paper, we overcome this limitation and propose a latent variable Fisher discriminant analysis model. We relax the instance-level labeling into bag-level, is a kind of semi-supervised (video-level labels of event type are required for semantic frame extraction) and incorporates a data-driven prior over the latent variables. Hence, our method combines the latent variable inference and dimension reduction in an unified bayesian framework. We test our method on MUSK and Corel data sets and yield competitive results compared to the baseline approach. We also demonstrate its capacity on the challenging TRECVID MED11 dataset for semantic keyframe extraction and conduct a human-factors ranking-based experimental evaluation, which clearly demonstrates our proposed method consistently extracts more semantically meaningful keyframes than challenging baselines.

High-throughput sequencing technology applied to cellular mRNA (RNA-Seq) has revolutionized transcriptome studies [19, 17, 35, among many others]. In contrast to microarray platforms, which it has replaced in many applications, RNA-Seq can not only be used to accurately quantify known transcripts, but also to reveal the precise structure of transcripts at single-nucleotide resolution. RNA-Seq based transcript reconstruction has therefore become a valuable tool for the completion of genome annotations [22, 11, for instance] and further enabled subsequent analyses of differentially expressed genes [2], transcript isoforms [6, 4] and exons [3], all of which generally rely on correctly inferred transcript inventories. De novo transcript reconstruction is thus a pivotal step in the analysis of RNA-Seq data. There are two conceptually different strategies to approach this problem: one can either assemble transcripts directly from RNA-Seq reads using methodology that originated from genome assembly approaches [13, 23, 25].

This work proposes a framework using temporal data and domain knowledge in order to analyze complex agronomical features. The expertise is first formalized in an ontology, under the form of concepts and relationships between them, and then used in conjunction with raw data and mathematical models to design a software sensor. Next the software sensor outputs are put in relation to product quality, assessed by quantitative measurements. This requires the use of advanced data analysis methods, such as functional regression. The methodology is applied to a case study involving an experimental design in French vineyards. The temporal data consist of sap flow measurements, and the goal is to explain fruit quality (sugar concentration and weight), using vine's water courses through the various vine phenological stages. The results are discussed, as well as the method genericity and robustness.

We formulate a finite axiomatization of the implication problem for inclusion and conditional independence atoms (dependencies) in the dependence logic context. The input of this problem is given by a finite set Σ {φ} consisting of conditional independence atoms and inclusion atoms, and the question to decide is whether the following logical consequence holds Σ φ. (1) Independence logic [12] and inclusion logic [6] are recent variants of dependence logic the semantics of which are defined over sets of assigments (teams) rather than a single assignment as in first-order logic.

The combination of multiple classifiers using ensemble methods is increasingly important for making progress in a variety of difficult prediction problems. We present a comparative analysis of several ensemble methods through two case studies in genomics, namely the prediction of genetic interactions and protein functions, to demonstrate their efficacy on real-world datasets and draw useful conclusions about their behavior. These methods include simple aggregation, meta-learning, cluster-based meta-learning, and ensemble selection using heterogeneous classifiers trained on resampled data to improve the diversity of their predictions. We present a detailed analysis of these methods across 4 genomics datasets and find the best of these methods offer statistically significant improvements over the state of the art in their respective domains. In addition, we establish a novel connection between ensemble selection and meta-learning, demonstrating how both of these disparate methods establish a balance between ensemble diversity and performance.

We informally call a stochastic process learnable if it admits a generalization error approaching zero in probability for any concept class with finite VC-dimension (IID processes are the simplest example). A mixture of learnable processes need not be learnable itself, and certainly its generalization error need not decay at the same rate. In this paper, we argue that it is natural in predictive PAC to condition not on the past observations but on the mixture component of the sample path. This definition not only matches what a realistic learner might demand, but also allows us to sidestep several otherwise grave problems in learning from dependent data. In particular, we give a novel PAC generalization bound for mixtures of learnable processes with a generalization error that is not worse than that of each mixture component. We also provide a characterization of mixtures of absolutely regular ($\beta$-mixing) processes, of independent probability-theoretic interest.

The nature has inspired several metaheuristics, outstanding among these is Ant Colony Optimization (ACO), which have proved to be very effective and efficient in problems of high complexity (NP-hard) in combinatorial optimization. This paper describes the implementation of an ACO model algorithm known as Elitist Ant System (EAS), applied to a combinatorial optimization problem called Job Shop Scheduling Problem (JSSP). We propose a method that seeks to reduce delays designating the operation immediately available, but considering the operations that lack little to be available and have a greater amount of pheromone. The performance of the algorithm was evaluated for problems of JSSP reference, comparing the quality of the solutions obtained regarding the best known solution of the most effective methods. The solutions were of good quality and obtained with a remarkable efficiency by having to make a very low number of objective function evaluations.

This paper introduces the concept of online "Semantic Advertising", which we see as the technology that will help realize the full potential of Internet advertising. Internet advertising is a rapidly growing and arguably a dominant form of advertising. A recent IDC report (Weide, 2013) estimates that the total Internet advertising spend in 2011 was 87.4 billion dollars ($35B for the U.S. only), and predicts an annual growth rate of 16% over the next 5 years. We argue that Semantic Advertising, (SA), enables us to address the challenge of delivering relevance at scale in Internet Advertising. Our argument is based on our work as a company developing semantic technology for better online advertising. Semantic technology (Hitzler, Krotzsch and Rudolph, 2009) can be described as algorithms and software that enable representation and reasoning based on meaning. Several companies such as Google, Microsoft and Yahoo, and smaller startup companies have developed semantic technologies for advertising.

HOL(y)Hammer is an online AI/ATP service for formal (computer-understandable) mathematics encoded in the HOL Light system. The service allows its users to upload and automatically process an arbitrary formal development (project) based on HOL Light, and to attack arbitrary conjectures that use the concepts defined in some of the uploaded projects. For that, the service uses several automated reasoning systems combined with several premise selection methods trained on all the project proofs. The projects that are readily available on the server for such query answering include the recent versions of the Flyspeck, Multivariate Analysis and Complex Analysis libraries. The service runs on a 48-CPU server, currently employing in parallel for each task 7 AI/ATP combinations and 4 decision procedures that contribute to its overall performance. The system is also available for local installation by interested users, who can customize it for their own proof development. An Emacs interface allowing parallel asynchronous queries to the service is also provided. The overall structure of the service is outlined, problems that arise and their solutions are discussed, and an initial account of using the system is given.