Statistical learning theory

About: Statistical learning theory is a research topic. Over the lifetime, 1618 publications have been published within this topic receiving 158033 citations.

Location-aware Computing Based on Support Vector Machine

Abstract: Support vector machine is a kind of machine learning algorithm based on statistical learning theory which mainly researches the learning of limited number of samples. Support vector machine takes a compromise between complexity of model and learning capability for best generalization. It settles the over-training problem well. This paper applies the method in the second phase of location fingerprinting technique. By using the dataset collected in the data-collecting phase, it establishes a model that takes received signal strengths as input and coordinates of the location as output. At last, test samples are used to test this model, the result proved the learning capability of support vector machine.

A Principle of Least Action for the Training of Neural Networks

Abstract: Neural networks have been achieving high generalization performance on many tasks despite being highly over-parameterized. Since classical statistical learning theory struggles to explain this behavior, much effort has recently been focused on uncovering the mechanisms behind it, in the hope of developing a more adequate theoretical framework and having a better control over the trained models. In this work, we adopt an alternate perspective, viewing the neural network as a dynamical system displacing input particles over time. We conduct a series of experiments and, by analyzing the network's behavior through its displacements, we show the presence of a low kinetic energy displacement bias in the transport map of the network, and link this bias with generalization performance. From this observation, we reformulate the learning problem as follows: finding neural networks which solve the task while transporting the data as efficiently as possible. This offers a novel formulation of the learning problem which allows us to provide regularity results for the solution network, based on Optimal Transport theory. From a practical viewpoint, this allows us to propose a new learning algorithm, which automatically adapts to the complexity of the given task, and leads to networks with a high generalization ability even in low data regimes.

A Data Fusion Approach Based on Parallel Support Vector Machine

Abstract: Support Vector Machine has some advantages, such as simple structure and good generalization, which is one implementation in Statistical Learning Theory. SVM offers a kind of effective way for the data fusion problem of little sample, non-linear and high dimension. In this paper, mobile agents are applied to data fusion system. The model and the study method of data fusion system are improved. An approach of data fusion based on SVM is proposed. The experiment results show that this hierarchical and parallel SVM training algorithm is efficient to deal with large-scale classification problems and has more satisfying accuracy in classification precision.

Improved SVR Model for Multi-Layer Buildup Factor Calculation

Abstract: Buildup factors are key parameters for the accuracy of the point kernel method, which is widely used for gamma ray dose rate calculations in shielding design and radiation safety analysis. Recently, a new compact mathematical model for buildup factor representation has been suggested for embedding into point kernel codes thus replacing traditionally generated complex mathematical expressions. The new regression model is based on Support Vector Machines (SVM) learning technique, which is an extension of Statistical Learning Theory. In this paper, a progress in an ongoing work on creating a compact and accurate model for multi-layer buildup factor representation is presented.