Showing 1 to 49 of 49 matching Articles
Results per page:
Export (CSV)
By
Hernández, Yasmín; Sucar, Luis Enrique; ArroyoFigueroa, Gustavo
2 Citations
We are developing an affective model for intelligent tutoring systems; thus, the tutor considers the affective state as well as the knowledge state of the student to give instruction to students. An important component of the affective model is the affective student model. This last one is rooted on the OCC cognitive model of emotions and the fivefactor model, and it is represented as a dynamic Bayesian network. The personality traits, goals and knowledge state are considered to establish the student affect. The affective model has been integrated to an intelligent learning environment for learning mobile robotics. We conducted an initial evaluation of the affective student model with a group of 20 under graduate and graduate students to evaluate the affective student model. Results are encouraging since they show a high agreement between the affective state established by the affective student model and the affective state reported by the students.
more …
By
MoralesGonzález, Annette; GarcíaReyes, Edel; Sucar, Luis Enrique
2 Citations
Image segmentation and Automatic Image Annotation (AIA) are two important areas that still impose challenging problems. Addressing both problems simultaneously may improve their results since they are interdependent. In this paper we give a step ahead in that direction considering different segmentation levels simultaneously and possible contextual relations among segments in order to improve the automatic image annotation. We propose to include hierarchical relations among regions of an image in a Markov Random Field (MRF) model for annotation. This relations are obtained from irregular pyramids, which keep parentchild relations among regions through all the levels. Our main contribution is therefore the combination of the irregular pyramid approach with context modeling by means of hierarchical MRFs. Experiments run in a subset of the Corel image collection showed a relevant improvement in the annotation accuracy.
more …
By
Sucar, Luis Enrique
9 Citations
No abstract available
By
Sucar, Luis Enrique
9 Citations
No abstract available
By
Díaz de León, Rocío; Sucar, Luis Enrique
We propose a general model for visual recognition of human activities, based on a probabilistic graphical framework. The motion of each limb and the coordination between them is considered in a layered network that can represent and recognize a wide range of human activities. By using this model and a sliding window, we can recognize simultaneous activities in a continuous way. We explore two inference methods for obtaining the most probable set of activities per window: probability propagation and abduction. In contrast with the standard approach that uses several models, we use a single classifier for multiple activity recognition. We evaluated the model with real image sequences of 6 different activities performed continuously by different people. The experiments show high recall and recognition rates.
more …
By
Morán, Alberto L.; Meza, Victoria; RamírezFernández, Cristina; Grimaldo, Ana I.; GarcíaCanseco, Eloísa; OrihuelaEspina, Felipe; Sucar, Luis Enrique
Show all (7)
4 Citations
We report the results of an indirect observation usability and user experience (UX) study on the use of the Gesture Therapy (GT) rehabilitation platform, as a physical activation and cognitive stimulation tool for the elderly. The results from this study complement those of a former selfreport study [8]. Elders perceived the system with high usefulness, usability, and UX, as well as generating low anxiety in both studies. Also, the results allowed us to analyze and evaluate the impact of elders’ previous experience on computer use on specific aspects. Interestingly, the significance of the effect of previous computer use experience on perceived anxiety and perceived enjoyment aspects of UX was different in both studies, although there is an important overlap for ease of use factors. These results, although not conclusive yet on the causes for the difference, provides us with further evidence to establish that elders’ previous experience (or not) on computer use affects their user experience on the use of the GT platform.
more …
By
MoralesGonzález, Annette; GarcíaReyes, Edel; Sucar, Luis Enrique
2 Citations
Image Segmentation and Automatic Image Annotation are two research fields usually addressed independently. Treating these problems simultaneously and taking advantage of each other’s information may improve their individual results. In this work our ultimate goal is image annotation, which we perform using the hierarchical structure of irregular pyramids. We propose a new criterion to create new segmentation levels in the pyramid using lowlevel cues and semantic information coming from the annotation step. Later, we use the improved segmentation to obtain better annotation results in an iterative way across the hierarchy.We perform experiments in a subset of the Corel dataset, showing the relevance of combining both processes to improve the results of the final annotation.
more …
By
Martínez, Miriam; Sucar, Luis Enrique; Acosta, Hector Gabriel; Cruz, Nicandro
Show all (4)
1 Citations
We have developed a novel methodology to combine several models using a Bayesian approach. The method selects the most relevant attributes from several models, and produces a Bayesian classifier which has a higher classification rate than any of them, and at the same time is very efficient. Based on conditional information measures, the method eliminates irrelevant variables, and joins or eliminates dependent variables; until an optimal Bayesian classifier is obtained. We have applied this method for diagnosis of precursor lesions of cervical cancer. The temporal evolution of the color changes in a sequence of colposcopy images is analyzed, and the resulting curve is fit to an approximate model. In previous work we develop 3 different mathematical models to describe the temporal evolution of each image region, and based on each model to detect regions that could have cancer. In this paper we combine the three models using our methodology and show very high accurracy for cancer detection, superior to any of the 3 original models.
more …
By
Sucar, Luis Enrique
9 Citations
No abstract available
By
Sucar, Luis Enrique
9 Citations
No abstract available
By
Sucar, Luis Enrique
9 Citations
No abstract available
By
Navarrete, Dulce J.; Morales, Eduardo F.; Sucar, Luis Enrique
Object recognition from images is traditionally based on a large training set of previously annotated images which is impractical for some applications. Also, most methods use only local or global features. Due to the nature of objects some features are better suited for some objects, so researchers have recently combined both types of features to improve the recognition performance. This approach, however, is not sufficient for the recognition of generic objects which can take a wide variety of appearances. In this paper, we propose a novel object recognition system that: (i) uses a small set of images obtained from the Web, (ii) induces a set of models for each object to deal with polymorphism, and (iii) optimizes the contribution of local and global features to deal with different types of objects. We performed tests with both generic and specific objects, and compared the proposed approach against base classifiers and stateoftheart systems with very promising results.
more …
By
Sucar, Luis Enrique
9 Citations
No abstract available
By
Arellano, Gerardo; Sucar, Luis Enrique; Morales, Eduardo F.
Automatic image annotation refers to the process of automatically labeling an image with a predefined set of keywords. Image annotation is an important step of contentbased image retrieval (CBIR), which is relevant for many realworld applications. In this paper, a new algorithm based on multiple grid segmentation, entropybased information and a Bayesian classifier, is proposed for an efficient, yet very effective, image annotation process. The proposed approach follows a two step process. In the first step, the algorithm generates grids of different sizes and different overlaps, and each grid is classified with a Naive Bayes classifier. In a second step, we used information based on the predicted class probability, its entropy, and the entropy of the neighbors of each grid element at the same and different resolutions, as input to a second binary classifier that qualifies the initial classification to select the correct segments. This significantly reduces false positives and improves the overall performance. We performed several experiments with images from the MSRC9 database collection, which has manual ground truth segmentation and annotation information. The results show that the proposed approach has a very good performance compared to the initial labeling, and it also improves other scheme based on multiple segmentations.
more …
By
Cabanillas, Javier; Morales, Eduardo F.; Sucar, Luis Enrique
2 Citations
Searching for an object in an environment using a mobile robot is a challenging task that requires an algorithm to define a set of points in which to sense the environment and an effective traversing strategy, to decide the order in which to visit such points. Previous work on sensing strategies normally assume unrealistic conditions like infinite visibility of the sensors. This paper introduces the concept of recognition area that considers the robot’s perceptual limitations. Three new sensing algorithms using the recognition area are proposed and tested over 20 different maps of increasing difficulty and their advantages over traditional algorithms are demonstrated. For the traversing strategy, a new heuristic is defined that significantly reduces the branching factor of a modified Branch & Bound algorithm, producing paths which are not too far away from the optimal paths but with several orders of magnitude faster that a traditional Branch & Bound algorithm.
more …
By
SolanoSoto, Jaime; Sucar, Luis Enrique
2 Citations
A novel configuration method for systems design has been developed, that considers, at the same time, system reliability and cost. This method helps to maximize the reliability, minimize the cost and obtain the best possible configuration for the system to be designed. To accomplish this, a combination of Bayesian networks and heuristic search are used so to help the designer find the optimum configuration in the immense search space available. The method has as entry parameters: the minimal reliability requirement or maximum cost of the computer system to be designed, the function of the system as a reliability block diagram and a description of each component. From this input, the methodology transforms automatically the reliability block diagram to Bayesian network equivalent, from which the reliability of the system is obtained through probability propagation. Starting form the initial block diagram, a set of heuristic operators is used to generate new configurations. The “best” configurations are obtained using beam search with some heuristics to improve the search efficiency. There are 3 alternatives for defining the best configurations: (i) minimize cost with a reliability restriction, (ii) maximize reliability with a cost restriction, and (iii) make a compromise between reliability and cost (Pareto set). The methodology is applied to the design of a distributed control system with promising results.
more …
By
Meléndez, Augusto; Sucar, Luis Enrique; Morales, Eduardo F.
2 Citations
Several methods have been developed for face detection with certain success, however these tend to fail under difficult conditions such as partial occlusions and changes in orientation and illumination. We propose a novel technique for face detection based on a visual grammar. We define a symbol relational grammar for faces, representing the visual elements of a face and their spatial relations. This grammar is transformed into a Bayesian network (BN) structure and its parameters are obtained from data, i.e., positive and negative examples of faces. Then the BN is used for face detection via probabilistic inference, using as evidence a set of weak detectors for different face components. We evaluated our method on a set of sample images of faces under difficult conditions, and contrasted it with a simplified model without relations, and the AdaBoost face detector. The results show a significant improvement when using our method.
more …
By
Ruiz, Elias; Sucar, Luis Enrique
2 Citations
A novel proposal for a general model for object recognition is presented. The proposed method is based on symbolrelational grammars and Bayesian networks. An object is modeled as a hierarchy of features and spatial relationships using a symbolrelational grammar. This grammar is learned automatically from examples, incorporating a simple segmentation algorithm in order to generate the lexicon. The grammar is created with the elements of the lexicon as terminal elements. This representation is automatically transformed into a Bayesian network structure which parameters are learned from examples. Thus, recognition is based on probabilistic inference in the Bayesian network representation. Preliminary results in modeling natural objects are presented. The main contribution of this work is a general methodology for building object recognition systems which combines the expressivity of a grammar with the robustness of probabilistic inference.
more …
By
Borrego, Gilberto; Morán, Alberto L.; LaFlor, Arturo; Meza, Victoria; GarcíaCanseco, Eloísa; OrihuelaEspina, Felipe; Sucar, Luis Enrique
Show all (7)
In this paper the design, development and preliminary evaluation of a serious videogame for the motor rehabilitation of upper limb and cognitive stimulation of the elderly are presented. The game includes features that allow (i) performing collaborative therapy exercises between two patients, (ii) remote configuration of the session therapy, and (iii) monitoring/analyzing of the session results by the therapist. A pilot evaluation with 7 older adults and an expert therapist, suggest that the game is perceived as stimulating, useful, usable and even funny, while providing an effective way to support/monitor the patient, and to adjust the therapy programs.
more …
By
Davis, Jesse; Sucar, Luis Enrique; OrihuelaEspina, Felipe
Treatment is the care and management of a patient to combat, ameliorate, or prevent a disease, disorder, or injury.
By
Heyer, Patrick; OrihuelaEspina, Felipe; Castrejón, Luis R.; HernándezFranco, Jorge; Sucar, Luis Enrique
Show all (5)
Given its virtually algorithmic process, the FuglMeyer Assessment (FMA) of motor recovery is prone to automatization reducing subjectivity, alleviating therapists’ burden and collaterally reducing costs. Several attempts have been recently reported to achieve such automatization of the FMA. However, a costeffective solution matching expert criteria is still unfulfilled, perhaps because these attempts are sensorspecific representation of the limb or have thus far rely on a trial and error strategy for building the underpinning computational model. Here, we propose a sensor abstracted representation. In particular, we improve previously reported results in the automatization of FMA by classifying a manifold embedded representation capitalizing on quaternions, and explore a wider range of classifiers. By enhancing the modeling, overall classification accuracy is boosted to 87% (mean: 82% ± 4.53:) well over the maximum reported in literature thus far 51.03% (mean: 48.72 ± std: 2.10). The improved model brings automatic FMA closer to practical usage with implications for rehabilitation programs both in ward and at home.
more …
By
HernándezGracidas, Carlos Arturo; Sucar, Luis Enrique; MontesyGómez, Manuel
8 Citations
In this paper we proposed the use of spatial relations as a way of improving annotationbased image retrieval. We analyzed different types of spatial relations and selected the most adequate ones for image retrieval. We developed an image comparison and retrieval method based on conceptual graphs, which incorporates spatial relations. Additionally, we proposed an alternative termweighting scheme and explored the use of more than one sample image for retrieval using several late fusion techniques. Our methods were evaluated with a rich and complex image dataset, based on the 39 topics developed for the ImageCLEF 2008 photo retrieval task. Results show that: (i) incorporating spatial relations produces a significant increase in performance, (ii) the label weighting scheme we proposed obtains better results than other traditional schemes, and (iii) the combination of several sample images using late fusion produces an additional improvement in retrieval according to several metrics.
more …
By
Chang, Leonardo; Duarte, Miriam Monica; Sucar, Luis Enrique; Morales, Eduardo F.
Show all (4)
1 Citations
Several methods have been presented in the literature that successfully used SIFT features for object identification, as they are reasonably invariant to translation, rotation, scale, illumination and partial occlusion. However, they have poor performance for classification tasks. In this work, SIFT features are used to solve problems of object class recognition in images using a twostep process. In its first step, the proposed method performs clustering on the extracted features in order to characterize the appearance of classes. Then, in the classification step, it uses a three layer Bayesian network for object class recognition. Experiments show quantitatively that clusters of SIFT features are suitable to represent classes of objects. The main contributions of this paper are the introduction of a Bayesian network approach in the classification step to improve performance in an object class recognition task, and a detailed experimentation that shows robustness to changes in illumination, scale, rotation and partial occlusion.
more …
By
TorresToledano, José Gerardo; Sucar, Luis Enrique
37 Citations
This paper presents an extension of Bayesian networks (BN) applied to reliability analysis. We developed a general methodology for reliability modeling of complex systems based on Bayesian networks. A reliability structure represented as a reliability block diagram is transformed to a Bayesian network representation, and with this, the reliability of the system can be obtained using probability propagation techniques. This allows for modeling complex systems, such as a bridge type, and dependencies between failures, which are difficult to obtain with conventional reliability analysis techniques. The relation between a BN and fault tree, and some advantages of BN for modeling system reliability are shown. We present some examples of the application of this methodology in solving difficult cases, which occur in reliability analysis of power plants.
more …
By
Sucar, Luis Enrique
1 Citations
This chapter covers Bayesian classifiers. After a brief introduction to the classification problem, the Naive Bayesian classifier is presented, as well as its main variants: TAN and BAN. Then the semiNaive Bayesian classifier is described. A multidimensional classifier may assign several classes to the same object. Two alternatives for multidimensional classification are analyzed: the multidimensional Bayesian network classifier and the Bayesian chain classifier. Then an introduction to hierarchical classification is presented. The chapter concludes by illustrating the application of Bayesian classifiers in two domains: skin pixel detection in images and drug selection for HIV treatment.
more …
By
Sucar, Luis Enrique
This chapter introduces sequential decision problems, in particular Markov decision processes (MDPs). A formal definition of an MDP is given, and the two most common solution techniques are described: value iteration and policy iteration. Then, factored MDPs are described, which provide a representation based on graphical models to solve very large MDPs. An introduction to partially observable MDPs (POMDPs) is also included. The chapter concludes by describing two applications of MDPs: power plant control and service robot task coordination.
more …
By
Sucar, Luis Enrique
1 Citations
In this chapter, a review of some aspects of graph theory that are important for probabilistic graphical models are presented. After providing a definition of directed and undirected graphs, some basic theoretical graph concepts are introduced, including types of graphs, trajectories and circuits, and graph isomorphism. A section is dedicated to trees, an important type of graph. Some more advanced theoretical graph aspects required for inference in probabilistic models are introduced, including cliques, triangulated graphs, and perfect orderings. The chapter concludes with a description of the maximum cardinality search and graph filling algorithms.
more …
By
Sucar, Luis Enrique
This chapter introduces Bayesian networks, covering representation and inference. The basic representational aspects of a Bayesian network are presented, including the concept of DSeparation and the independence axioms. With respect to parameter specification, the two main alternatives for a compact representation are described, one based on canonical models and the other on graphical representations. Then the main algorithms for probabilistic inference are introduced, including belief propagation, variable elimination, conditioning, junction trees, loopy propagation, and stochastic simulation. The chapter concludes by illustrating the application of Bayesian networks in information validation and system reliability analysis.
more …
By
Sucar, Luis Enrique
9 Citations
No abstract available
By
Díaz de León, Rocío; Sucar, Luis Enrique
Abstract
We propose a general model for visual recognition of human activities, based on a probabilistic graphical framework. The motion of each limb and the coordination between them is considered in a layered network that can represent and recognize a wide range of human activities. By using this model and a sliding window, we can recognize simultaneous activities in a continuous way. We explore two inference methods for obtaining the most probable set of activities per window: probability propagation and abduction. In contrast with the standard approach that uses several models, we use a single classifier for multiple activity recognition. We evaluated the model with real image sequences of 6 different activities performed continuously by different people. The experiments show high recall and recognition rates.
more …
By
Ibargüengoytia, Pablo H.; Reyes, Alberto; RomeroLeon, Inés; Pech, David; García, Uriel A.; Sucar, Luis Enrique; Morales, Eduardo F.
Show all (7)
3 Citations
This paper presents the development of a novel dynamic Bayesian network (DBN) model devoted to wind forecasting. An original procedure was developed to approximate this model, based on historical information in the form of time series. The DBN structure and parameters are learned from historical data, and this methodology can be applied to any prediction problem. In contrast to previous approaches, the proposed model considers all the relevant variables in the domain and produces a probability distribution for the predictions; providing important additional information to the decision makers. The method was evaluated experimentally with real data from a wind farm in Mexico for a time horizon of 5 hours, showing superior performance to traditional timeseries prediction techniques.
more …
By
Reyes, Alberto; Ibargüengoytia, Pablo H.; Sucar, Luis Enrique
1 Citations
Markov decision processes (MDPs) provide a powerful framework for solving planning problems under uncertainty. However, it is difficult to apply them to real world domains due to complexity and representation problems: (i) the state space grows exponentially with the number of variables; (ii) a reward function must be specified for each stateaction pair. In this work we tackle both problems and apply MDPs for a complex real world domain combined cycle power plant operation. For reducing the state space complexity we use a factored representation based on a two–stage dynamic Bayesian network [13]. The reward function is represented based on the recommended optimal operation curve for the power plant. The model has been implemented and tested with a power plant simulator with promising results.
more …
By
Galán, Severino F.; ArroyoFigueroa, Gustavo; Díez, F. Javier; Sucar, Luis Enrique
Show all (4)
1 Citations
Temporal Nodes Bayesian Networks (TNBNs) and Networks of Probabilistic Events in Discrete Time (NPEDTs) are two different types of Bayesian networks (BNs) for temporal reasoning. ArroyoFigueroa and Sucar applied TNBNs to an industrial domain: the diagnosis and prediction of the temporal faults that may occur in the steam generator of a fossil power plant. We have recently developed an NPEDT for the same domain. In this paper, we present a comparative evaluation of these two systems. The results show that, in this domain, NPEDTs perform better than TNBNs. The ultimate reason for that seems to be the finer time granularity used in the NPEDT with respect to that of the TNBN. Since families of nodes in a TNBN interact through the general model, only a small number of states can be defined for each node; this limitation is overcome in an NPEDT through the use of temporal noisy gates.
more …
By
Sucar, Luis Enrique
This chapter presents an overview of some basic concepts in probability theory which are important for understanding probabilistic graphical models. First, the main interpretations and mathematical definition of probability are introduced. Second, the basic rules of probability theory are presented, including the concept of conditional independence and Bayes rule. Third, an overview of random variables and some important distributions are described. Lastly, the basics of information theory are presented.
more …
By
Sucar, Luis Enrique
This chapter presents an introduction to Markov random fields (MRFs), also known as Markov networks, which are undirected graphical models. We describe how a Markov random field is represented, including its structure and parameters, with emphasis on regular MRFs. Then, a general stochastic simulation algorithm to find the optimum configuration of an MRF is described, including some of its main variants. The problem of parameter estimation for an MRF is addressed, considering the maximum likelihood estimator. Conditional random fields are also introduced. The chapter concludes with two applications of MRFs for image analysis, one for image denoising and the other for improving image annotation by including spatial relations.
more …
By
Sucar, Luis Enrique
2 Citations
This chapter introduces relational probabilistic graphical models (RPGMs), which combine the expressive power of predicate logic with the uncertain reasoning capabilities of probabilistic graphical models. First, a brief review of propositional and predicate logic is presented. Then, two different relational probabilistic formalisms are described: probabilistic relational models and Markov logic networks. Finally, the application of the two previous approaches is illustrated in two domains, student modeling for a virtual laboratory and visual object recognition based on symbolrelational grammars.
more …
By
Sucar, Luis Enrique
Dynamic Bayesian network models extend BNs to represent the temporal evolution of a certain process. There are two basic types of Bayesian network models for dynamic processes: state based and event based. Dynamic Bayesian networks are statebased models that represent the state of each variable at discrete time intervals. Eventbased models represent the changes in the state of each state variable; each temporal variable will then correspond to the time in which a state change occurs. In this chapter, we will review dynamic Bayesian networks and event networks, including representation, inference, and learning. The chapter includes two application examples: dynamic Bayesian networks for gesture recognition and temporal nodes Bayesian networks for HIV mutational pathways prediction.
more …
By
Morán, Alberto L.; OrihuelaEspina, Felipe; MezaKubo, Victoria; Grimaldo, Ana I.; RamírezFernández, Cristina; GarcíaCanseco, Eloisa; OropezaSalas, Juan Manuel; Sucar, Luis Enrique
Show all (8)
5 Citations
We explore the use of a virtual rehabilitation platform as the interaction means for physical activation and cognitive stimulation of elders. A usability evaluation of actual and projected use of the tool suggests that this could be feasible to perform. Elders perceived the use of the evaluated tool as useful (93.75/100), easy to use (93.75/100) and pleasurable to use (91.66/100) during an actual activation and stimulation session. Previous experience on the use of computers by the participants did not significantly impact on their usability perception for most of the included factors, with the sole exception being the perception of anxiety. This is an encouraging result to reuse and adapt technologies from “close” domains (e.g., virtual rehabilitation). In addition, this can reduce development times and cost, and facilitate knowledge transfer into the domain of physical activation and cognitive stimulation of elders.
more …
By
Sucar, Luis Enrique
This chapter gives an introduction to causal modeling, in particular to causal Bayesian networks. It starts by introducing causal models and their importance. Then causal Bayesian networks are described, including two types of causal reasoning, prediction and counterfactuals. It continues with the topic of learning causal models, presenting one of the stateoftheart techniques. Finally, it shows an example of learning causal models from realworld data about children with Attention Deficit Hyperactivity Disorder.
more …
By
Arroyo, Jaen Alberto; GomezCastaneda, Cecilia; Ruiz, Elias; Cote, Enrique Munoz; Gavi, Francisco; Sucar, Luis Enrique
Show all (6)
This paper presents a method to assess nitrogen levels, a nitrogen nutrition index (NNI), in corn crops (Zea mays) using multispectral remote sensing imagery. The multispectral sensors used were four spectral bands only. The experiments were compared with nitrogen levels sensed in the field. The corn crops were divided into three nitrogen fertilization levels (70, 140 and 210
$$\mathrm{kg N}\cdot \mathrm{ha}^{1}$$
) into three replicates. In this sense, we propose a method to infer nitrogen levels in corn crops by using airborne multispectral sensors and machine learning techniques. The presented results offered a simple model to estimate nitrogen with lowcost technologies (UAVs and multispectral cameras only) in small to medium size areas of corn crops.
more …
By
Sucar, Luis Enrique
This introductory chapter starts by describing the effects of uncertainty in intelligent systems and presents a brief history of the development of uncertain reasoning in artificial intelligence. Then it presents the basic approach for probabilistic reasoning, motivating the development of probabilistic graphical models. It gives an overview of probabilistic graphical models, the types of models, and how these can be classified. It concludes with a description of the rest of the book.
more …
By
Chang, Leonardo; PérezSuárez, Airel; RodríguezCollada, Máximo; HernándezPalancar, José; AriasEstrada, Miguel; Sucar, Luis Enrique
Show all (6)
Bag of Visual Words is one of the most widely used approaches for representing images for object categorization; however, it has several drawbacks. In this paper, we propose three properties and their corresponding quantitative evaluation measures to assess the ability of a visual word to represent and discriminate an object class. Additionally, we also introduce two methods for ranking and filtering visual vocabularies and a soft weighting method for BoW image representation. Experiments conducted on the Caltech101 dataset showed the improvement introduced by our proposals, which obtained the best classification results for the highest compression rates when compared with a stateoftheart mutual information based method for feature selection.
more …
By
Reyes, Alberto; Ibargüengoytia, Pablo H.; Sucar, Luis Enrique
Markov decision processes (MDPs) provide a powerful framework for solving planning problems under uncertainty. However, it is difficult to apply them to real world domains due to complexity and representation problems: (i) the state space grows exponentially with the number of variables; (ii) a reward function must be specified for each stateaction pair. In this work we tackle both problems and apply MDPs for a complex real world domain combined cycle power plant operation. For reducing the state space complexity we use a factored representation based on a two–stage dynamic Bayesian network [13]. The reward function is represented based on the recommended optimal operation curve for the power plant. The model has been implemented and tested with a power plant simulator with promising results.
more …
By
Sucar, Luis Enrique
3 Citations
Markov chains and hidden Markov models (HMMs) are particular types of PGMs that represent dynamic processes. After a brief introduction to Markov chains, this chapter focuses on hidden Markov models. The algorithms for solving the basic problems: evaluation, optimal sequence, and parameter learning are presented. The chapter concludes with a description of several extensions to the basic HMM, and two applications: the “PageRank” procedure used by Google and gesture recognition.
more …
By
Sucar, Luis Enrique
This chapter gives an introduction to learning Bayesian networks including both parameter and structure learning. Parameter learning includes how to handle uncertainty in the parameters and missing data; it also includes the basic discretization techniques. After describing the techniques for learning tree and polytree BNs, the two main types of methods for structure learning are described: score and search, and independence tests. We then describe how to combine expert knowledge and data. The chapter concludes with an application example in the area of pollution modeling.
more …
By
Sucar, Luis Enrique; ÁvilaSansores, Shender Maria; OrihuelaEspina, Felipe
Intelligent rehabilitation is a novel paradigm in motor rehabilitation empowering assistive technology with artificial intelligence (AI). Central to this paradigm is adaptation, the capacity of the assistive technology to dynamically accommodate to the therapy evolving demands. This chapter overviews several existing AI solutions to implement a decision making model to provide rehabilitation tools with adaptation capabilities, and provides details of a powerful approach capable of exploiting prior knowledge for a quick start and posterior knowledge to guarantee uptodated informed decisions. In this solution, a Markov decision process formulates an initial policy optimal within prior knowledge; a policy which is later on allow to evolve on incoming evidence to fit new requirements. This solution ensures short training periods and exhibits convergence with therapists’ criteria. In consequence, intelligent adaptation to dynamic circumstances of the patient and therapy plan is demonstrated a feasible endeavour within a real practical timeline. This might endow assistive technology with the necessary competence to be taken home and/or reduce expert surpervision.
more …
By
Sucar, Luis Enrique
This chapter introduces decision models. First, a brief review of the fundamentals of decision theory is presented. Second, we describe decision trees and their evaluation strategy. Third, influence diagrams are introduced, including two alternative evaluation strategies: variable elimination and transformation to a Bayesian network. The chapter concludes with an application of a decision model that acts as a caregiver to guide an elderly or handicapped person in cleaning her hands.
more …
By
ArroyoFigueroa, Gustavo; Sucar, Luis Enrique
15 Citations
In some domains like industry, medicine, communications, speech recognition, planning, tutoring systems, and forecasting; the timing of observations (symptoms, measures, test, events, as well as faults) play a major role in diagnosis and prediction. This paper introduces a new formalism to deal with uncertainty and time using Bayesian networks called Temporal Bayesian Network of Events (TBNE). In a TBNE each node represents an event or state change of a variable, and an arc corresponds to a causaltemporal relationship. A temporal node represents the time that a variable changes state, including an option of nochange. The temporal intervals can differ in number and size for each temporal node, so this allows multiple granularity. Our approach is contrasted with a Dynamic Bayesian network for a simple medical example. An empirical evaluation is presented for a subsystem of a thermal power plant, in which this approach is used for fault diagnosis and event prediction with good results. The TBNE model can be used for the diagnosis of a cascade of anomalies arising with certain delays; this situation is typical in the diagnosis of medical and industrial processes.
more …
By
Sucar, Luis Enrique; PérezBrito, Joaquín; RuizSuárez, J. Carlos; Morales, Eduardo
Show all (4)
16 Citations
In this paper we propose an algorithm for structure learning in predictive expert systems based on a probabilistic network representation. The idea is to have the “simplest” structure (minimum number of links) with acceptable predictive capability. The algorithm starts by building a tree structure based on measuring mutual information between pairs of variables, and then it adds links as necessary to obtain certain predictive performance. We have applied this method for ozone prediction in México City, where the ozone level is used as a global indicator for the air quality in different parts of the city. It is important to predict the ozone level a day, or at least several hours in advance, to reduce the health hazards and industrial losses that occur when the ozone reaches emergency levels. We obtained as a first approximation a treestructured dependency model for predicting ozone in one part of the city. We observe that even with only three parameters, its estimations are acceptable.
A causal network representation and the structure learning techniques produced some very interesting results for the ozone prediction problem. Firstly, we got some insight into the dependence structure of the phenomena. Secondly, we got an indication of which are the important and not so important variables for ozone forecasting. Taking this into account, the measurement and computational costs for ozone prediction could be reduced. And thirdly, we have obtained satisfactory short term ozone predictions based on a small set of the most important parameters.
more …
