Deep learning has shown promising results in many machine learning applications. The hierarchical feature representation built by deep networks enable compact and precise encoding of the data. A kernel analysis of the trained deep networks demonstrated that with deeper layers, more simple and more accurate data representations are obtained. In this paper, we propose an approach for layer-wise t...

Hierarchical neural networks for object recognition have a long history. In recent years, novel methods for incrementally learning a hierarchy of features from unlabeled inputs were proposed as good starting point for supervised training. These deep learning methods— together with the advances of parallel computers—made it possible to successfully attack problems that were not practical before,...

Deep networks can potentially express a learning problem more efficiently than local learning machines. While deep networks outperform local learning machines on some problems, it is still unclear how their nice representation emerges from their complex structure. We present an analysis based on Gaussian kernels that measures how the representation of the learning problem evolves layer after la...

A new design concept is introduced to control the near-wall integration between the hot-gas boundary layer and the cooling jets in order to enhance the adiabatic film cooling effectiveness of the gas turbine blades. In this new approach, another film cooling port, having a very low blowing ratio, which prevents formation of the counter-rotating vortex pare, is applied just upstream of the main ...

While deep neural networks have succeeded in several visual applications, such as object recognition, detection, and localization, by reaching very high classification accuracies, it is important to note that many real-world applications demand varying costs for different types of misclassification errors, thus requiring cost-sensitive classification algorithms. Current models of deep neural ne...

In this paper, we present a layer-wise learning of stochastic neural networks (SNNs) in an information-theoretic perspective. In each layer of an SNN, the compression and the relevance are defined to quantify the amount of information that the layer contains about the input space and the target space, respectively. We jointly optimize the compression and the relevance of all parameters in an SN...

Recently, a successful pose estimation algorithm, called Cascade Pose Regression (CPR), was proposed in the literature. Trained over Pose Index Feature, CPR is a regressor ensemble that is similar to Boosting. In this paper we show how CPR can be represented as a Neural Network. Specifically, we adopt a Graph Transformer Network (GTN) representation and accordingly train CPR with Back Propagati...

This work proposes innovative permutation-based procedures controlling the Familywise Error rate (FWE). It is proofed that weighted procedures control the FWE if weights are a function of the sufficient statistic. Particularly, we focus on the use of the additional information given by the total variance of each variables. The first proposal considers the use of weights applied to the combining...

The advancement in wireless communication technologies is becoming more demanding and pervasive. One of the fundamental parameters that limit efficiency network are security challenges. vulnerable to attacks such as spoofing signal strength attacks. Intrusion detection signifies a central approach ensuring network. In this paper, an Detection System based on framework graph theory proposed. A L...