Gonzalo Nápoles

Two-steps learning of Fuzzy Cognitive Maps for prediction and knowledge discovery on the HIV-1 drug resistance

The Human Immunodeficiency Virus (HIV) is a complex retrovirus that progressively deteriorates the immune system of infected patients, eventually causing death. Although antiviral drugs are not able to eradicate the HIV, they are designed to inhibit the function of three essential proteins in the virus replication process: protease, reverse transcriptase and integrase. However, due to a high mutation rate, this virus is capable to develop resistance to existing drugs causing the treatment failure. Several machine learning techniques have been proposed for predicting HIV drugs resistance, but most of them are difficult to interpret. Actually, in last years the protein modeling of this virus has become, from diverse points of view, an open problem for researchers. In this paper we propose a model based on Fuzzy Cognitive Maps (FCM) for analyzing the behavior of the HIV-1 protease protein. With this goal in mind, a two-steps learning algorithm using Swarm Intelligence for optimizing the modeling parameters is introduced. The first step is oriented to estimate the biological causality among amino acids describing the sequence through a continuous search method. While resulting adjusted maps are combined into a single one through an aggregation procedure for obtaining an initial prototype map. The second step optimizes the prototype by finding those amino acids directly associated with resistance using a discrete meta-heuristic. At the end, a fully optimized prototype map is obtained allowing predicting HIV-1 drug resistance and also discovering relevant knowledge on causal influences directly associated with resistance, for seven well-known protease inhibitors.

On the convergence of sigmoid Fuzzy Cognitive Maps

Fuzzy Cognitive Maps (FCM) are Recurrent Neural Networks that are used for modeling complex dynamical systems using causal relations. Similarly to other recurrent models, a FCM-based system repeatedly propagates an initial activation vector over the causal network until either the map converges to a fixed-point or a maximal number of cycles is reached. The former scenario leads to a hidden pattern, whereas the latter implies that cyclic or chaotic configurations may be produced. It should be highlighted that FCM equipped with discrete transfer functions never exhibit chaotic states, but this premise cannot be ensured for systems having continuous neurons. Recently, a few studies dealing with convergence on continuous FCM have been introduced. However, such methods are not suitable for FCM-based systems used in pattern classification environments. In this paper, we first review a new heuristic procedure called Stability based on sigmoid Functions, which allows to improve the convergence on sigmoid FCM, without modifying the weights configuration. Afterwards, we examine some drawbacks that affect the algorithm performance and introduce solutions to enhance its performance in pattern classification environments. Additionally, we formalize several definitions which were omitted in the original research. These solutions lead to accurate classifiers and prevent specific scenarios where the method may fail. Towards the end, we conduct numerical simulations across six FCM-based classifiers with unstable features in order to evaluate the proposed improvements in pattern classification environments.

Rough Cognitive Networks

Decision-making could informally be defined as the process of selecting the most appropriate actions among a set of possible alternatives in a given activity. In recent years several decision models based on Rough Set Theory (e.g. three-way decision rules) and Fuzzy Cognitive Maps have been introduced for addressing such problems. However, most of them are focused on decision-making problems with discrete attributes or they are oriented to specific domains. In this paper we present a decision model called Rough Cognitive Networks that combines the abstract semantic of the three-way decision model with the neural reasoning mechanism of Fuzzy Cognitive Maps for addressing numerical decision-making problems. The contribution of this study is two-fold. On one hand, it allows to explicitly handle decision-making problems with numerical features, where the target object could activate multiple regions at the same time. On the other hand, in such granular networks the three-way decision rules are used to design the topology of the map, addressing in some sense the inherent limitations in the expression and architecture of Fuzzy Cognitive Maps. Moreover, we propose a learning methodology using Harmony Search for adjusting the model parameters, leading to a parameter-free decision model where the human intervention is not required. A comparative analysis with standard classifiers and recently proposed rough recognition models is conducted in order to show the effectiveness of the proposal.

Constricted Particle Swarm Optimization based Algorithm for Global Optimization

Particle Swarm Optimization (PSO) is a bioinspired meta-heuristic for solving complex global optimizat ion problems. In standard PSO, the particle swarm frequently gets attracted by suboptimal solutions, causing premature convergence of the algorithm and swarm stagnation. Once the particles have been attracted to a local optimum, they continue the sea rch process within a minuscule region of the solution space, an d escaping from this local optimum may be difficult. This pape r presents a modified variant of constricted PSO that uses rando m samples in variable neighborhoods for dispersing the swarm whenever a premature convergence (or stagnation) state is dete cted, offering an escaping alternative from local optima. The perf ormance of the proposed algorithm is discussed and experimental results show its ability to approximate to the global minimum in each of the nine well-known studied benchmark functions.

How to improve the convergence on sigmoid Fuzzy Cognitive Maps

Fuzzy Cognitive Maps (FCM) may be defined as Recurrent Neural Networks that allow causal reasoning. According to the transformation function used for updating the activation value of concepts they can be characterized as discrete or continuous. It is remarkable that FCM having discrete neurons never exhibit chaotic states, but this premise cannot be guaranteed for FCM having continuous concepts. On the other hand, complex Sigmoid FCM resulting from experts or learning algorithms often show chaotic or cyclic patterns, therefore leading to confusing interpretation of the investigated system. The first contribution of this paper is focused on explaining why most studies on FCM stability are not applicable to FCM used on classification or decision-making tasks. Next we describe a non-direct learning methodology based on Swarm Intelligence for improving the system stability once the causal weight estimation is done. The objective here is to find a specific threshold function for each map neuron simulating an external stimulus, instead of using the same transformation function for all concepts. At the end, we can compute more stable maps, so better consistency in hidden patterns is achieved.

Modelling, aggregation and simulation of a dynamic biological system through fuzzy cognitive maps

The complex dynamics of Human Immunodeficiency Virus leads to serious problems on predicting the drug resistance. Several machine learning techniques have been proposed for modelling this classification problem, but most of them are difficult to aggregate and interpret. In fact, in last years the protein modelling of this virus has become, from diverse points of view, an open problem for researchers. This paper presents a modelling of the protease protein as a dynamic system through Fuzzy Cognitive Maps, using the amino acids contact energies for the sequence description. In addition, a learning scheme based on swarm intelligence called PSO-RSVN is used to estimate the causal weight matrix that characterizes these structures. Finally, an aggregation procedure with previously adjusted maps is applied for obtaining a prototype map, in order to discover knowledge in the causal influences, and simulate the system behaviour when a single (or multiple) mutation takes place.

A fuzzy cognitive maps modeling, learning and simulation framework for studying complex system

This paper presents Fuzzy Cognitive Maps as an approach in modeling the behavior and operation of complex systems; they combine aspects of fuzzy logic, neural networks, semantic networks, expert systems, and nonlinear dynamical systems. They are fuzzy weighted directed graphs with feedback that create models that emulate the behavior of complex decision processes using fuzzy causal relations. First, the description and the methodology that this theory suggests is examined, later some ideas for using this approach in the control process area are discussed. An inspired on particle swarm optimization learning method for this technique is proposed, and then, the implementation of a tool based on Fuzzy Cognitive Maps is described. The application of this theory might contribute to the progress of more intelligent and independent systems. Fuzzy Cognitive Maps have been fruitfully used in decision making and simulation of complex situation and analysis. At the end, a case study about Travel Behavior is analyzed and results are assessed.

Hybrid Model Based on Rough Sets Theory and Fuzzy Cognitive Maps for Decision-Making

Decision-making could be defined as the process to choose a suitable decision among a set of possible alternatives in a given activity. It is a relevant subject in numerous disciplines such as engineering, psychology, risk analysis, operations research, etc. However, most real-life problems are unstructured in nature, often involving vagueness and uncertainty features. It makes difficult to apply exact models, being necessary to adopt approximate algorithms based on Artificial Intelligence and Soft Computing techniques. In this paper we present a novel decision-making model called Rough Cognitive Networks. It combines the capability of Rough Sets Theory for handling inconsistent patterns, with the modeling and simulation features of Fuzzy Cognitive Maps. Towards the end, we obtain an accurate hybrid model that allows to solve non-trivial continuous, discrete, or mixed-variable decision-making problems.

Learning Stability Features on Sigmoid Fuzzy Cognitive Maps through a Swarm Intelligence Approach

Fuzzy Cognitive Maps FCM are a proper knowledge-based tool for modeling and simulation. They are denoted as directed weighted graphs with feedback allowing causal reasoning. According to the transformation function used for updating the activation value of concepts, FCM can be grouped in two large clusters: discrete and continuous. It is notable that FCM having discrete outputs never exhibit chaotic states, but this premise can not be ensured for FCM having continuous output. This paper proposes a learning methodology based on Swarm Intelligence for estimating the most adequate transformation function for each map neuron concept. As a result, we can obtain FCM showing better stability properties, allowing better consistency in the hidden patterns codified by the map. The performance of the proposed methodology is studied by using six challenging FCM concerning the field of the HIV protein modeling.

Rough cognitive ensembles

Abstract Rough Cognitive Networks are granular classifiers stemming from the hybridization of Fuzzy Cognitive Maps and Rough Set Theory. Such cognitive neural networks attempt to quantify the impact of rough granular constructs (i.e., the positive, negative and boundary regions of a target concept) over each decision class for the problem at hand. In rough classifiers, determining the precise granularity level is crucial to compute high prediction rates. Regrettably, learning the similarity threshold parameter requires reconstructing the information granules, which may be time-consuming. In this paper, we put forth a new multiclassifier system classifier named Rough Cognitive Ensembles. The proposed ensemble employs a collection of Rough Cognitive Networks as base classifiers, each operating at a different granularity level. This allows suppressing the requirement of learning a similarity threshold. We evaluate the granular ensemble with 140 traditional classification datasets using different heterogeneous distance functions. After comparing the proposed model to 15 well-known classifiers, the experimental evidence confirms that our scheme yields very promising classification rates.