Catálogo de publicaciones - libros

A system that is capable of retaining learned knowledge and selectively transferring portions of that knowledge as a source of inductive bias during new learning would be a significant advance in artificial intelligence and inductive modeling. We define such a system to be a machine lifelong learning, or ML3 system. This paper makes an initial effort at specifying the scope of ML3 systems and their functional requirements.

In real life, the task learning is reinforced by the related tasks that we have learned or that we learn at the same time. This scheme applied to Artificial Neural Networks (ANN) is known with the name of Multitask Learning (MTL). So, the information coming from the related secondary tasks provide a bias to the main task, which improves its performances versus a Single-Task Learning (STL) scheme. However, this implies a bigger complexity. Data Editing procedures are used to reduce the algorithmic complexity, obtaining an outstanding samples set from the original set. This edited set gets the performance very fast. In this paper we combine MTL with Data Editing, so we can approach the small samples set training in an MTL scheme.

The goal of this work is to present an efficient implementation of the Backpropagation (BP) algorithm to train Artificial Neural Networks with general feedforward topology. This will lead us to the “consecutive retrieval problem” that studies how to arrange efficiently sets into a sequence so that every set appears contiguously in the sequence. The BP implementation is analyzed, comparing efficiency results with another similar tool. Together with the BP implementation, the data description and manipulation features of our toolkit facilitates the development of experiments in numerous fields.

Support Vector Machines (SVMs) with few support vectors are quite desirable, as they have a fast application to new, unseen patterns. In this work we shall study the coefficient structure of the dual representation of SVMs constructed for nonlinearly separable problems through kernel perceptron training. We shall relate them with the margin of their support vectors (SVs) and also with the number of iterations in which these SVs take part. These considerations will lead to a remove–and–retrain procedure for building SVMs with a small number of SVs where both suitably small and large coefficient SVs will be taken out from the training sample. Besides providing a significant SV reduction, our method’s computational cost is comparable to that of a single SVM training.

Existing algorithms for learning Bayesian network require a lot of computation on high dimensional itemsets which affects reliability, robustness and accuracy of these algorithms and takes up a large amount of time. To address the above problem, we propose a new Bayesian network learning algorithm MRMRG, Max Relevance-Min Redundancy Greedy. MRMRG algorithm is a variant of K2 which is a well-known BN learning algorithm. We also analyze the time complexity of MRMRG. The experimental results show that MRMRG algorithm has much better efficiency. It is also shown that MRMRG algorithm has better accuracy than most of existing learning algorithms for limited sample datasets.

Psychological studies have shown that emotion and affect influence learning. We employ these findings in a machine-learning meta-parameter context, and dynamically couple an adaptive agent’s artificial affect to its action-selection mechanism (Boltzmann ). The agent’s performance on two important learning problems is measured. The first consists of learning to cope with two alternating goals. The second consists of learning to prefer a later larger reward (global optimum) for an earlier smaller one (local optimum). Results show that, compared to several control conditions, coupling positive affect to exploitation and negative affect to exploration has several important benefits. In the alternating-goal task, it significantly reduces the agent’s “goal-switch search peak”. The agent finds its new goal faster. In the second task, artificial affect facilitates convergence to a global instead of a local optimum, while permitting to exploit that local optimum. We conclude that affect-controlled action-selection has adaptation benefits.

One of the main challenges in the information security concerns the introduction of systems able to identify intrusions. In this ambit this work takes place describing a new Intrusion Detection System based on anomaly approach. We realized a system with a hybrid solution between host-based and network-based approaches, and it consisted of two subsystems: a statistical system and a neural one. The features extracted from the network traffic belong only to the IP Header and their trend allows us detecting through a simple visual inspection if an attack occurred. Really the two-tier neural system has to indicate the status of the system. It classifies the traffic of the monitored host, distinguishing the background traffic from the anomalous one. Besides, a very important aspect is that the system is able to classify different instances of the same attack in the same class, establishing which attack occurs.

This work presents a noise cancellation system suitable for different biomedical signals based on a multilayer artifical neural network(ANN). The proposed method consists of a simple structure similar to the MADALINE neuronal network (Multiple ADAptive LINear Element). This network is a grown artificial neuronal network which allows to optimize the number of nodes of one hidden layer and coefficients of several matrixes. These coefficients matrixes are optimized using the Widrow-Hoff Delta algorithm which requires smaller computational cost than the required by the back-propagation algorithm.

The method’s performance has been obtained by computing the cross correlation between the input and the output signals to the system. In addition, the signal to interference ratio (SIR) has also been computed. Making use of the aforementioned indexes it has been possible to compare the different classical methods (Filter FIR, biorthogonal Wavelet 6,8, Filtered Adaptive LMS) and the proposed system based on neural multilayer networks . The comparison shows that the ANN-based method is able to better preserve the signal waveform at system output with an improved noise reduction in comparison with traditional techniques. Moreover, the ANN technique is able to reduce a great variety of noise signals present in biomedical recordings, like high frequency noise, white noise, movement artifacts and muscular noise.

On chip resource distribution is a problem that, due to its complexity, is susceptible to be solved by using artificial intelligence optimization procedures. In this paper, a Hopfield recurrent neural network and a Boltzmann machine are proposed for searching good solutions.

The main challenge of this approach is proposing an energy function to be minimized so it mixes all the problem-related restrictions.

Experimental data shows that we can get good enough solutions in a reasonable time using Hopfield nets or close to the global minimum solutions using Boltzmann machines.

Finding appropriate features is one of the key problems in the increasing applications of rough set theory, which is also one of the bottlenecks of the rough set methodology. Particle Swarm Optimization (PSO) is particularly attractive for this challenging problem. In this paper, we attempt to solve the problem using a particle swarm optimization approach. The proposed approach discover the best feature combinations in an efficient way to observe the change of positive region as the particles proceed through the search space. We evaluate the performance of the proposed PSO algorithm with Genetic Algorithm (GA). Empirical results indicate that the proposed algorithm could be an ideal approach for solving the feature reduction problem when other algorithms failed to give a better solution.