Results 1 - 10
of
36
Automated synthesis of body schema using multiple sensor modalities
- In Proceedings of the Tenth International Conference on the Simulation and Synthesis of Living Systems (ALIFEX
, 2006
"... The way in which organisms create body schema, based on their interactions with the real world, is an unsolved problem in neuroscience. Similarly, in evolutionary robotics, a robot learns to behave in the real world either without recourse to an internal model (requiring at least hundreds of interac ..."
Abstract
-
Cited by 11 (0 self)
- Add to MetaCart
(Show Context)
The way in which organisms create body schema, based on their interactions with the real world, is an unsolved problem in neuroscience. Similarly, in evolutionary robotics, a robot learns to behave in the real world either without recourse to an internal model (requiring at least hundreds of interactions), or a model is hand designed by the experimenter (requiring much prior knowledge about the robot and its environment). In this paper we present a method that allows a physical robot to automatically synthesize a body schema, using multimodal sensor data that it obtains through interaction with the real world. Furthermore, this synthesis can be either parametric (the experimenter provides an approximate model and the robot then refines the model) or topological: the robot synthesizes a predictive model of its own body plan using little
Action-selection and crossover strategies for self-modeling machines
- in Proc. Genetic Evol. Comput. Conf. (GECCO’07), 2007
"... In previous work [7] a computational framework was demonstrated that employs evolutionary algorithms to automatically model a given system. This is accomplished by alternating the evolution of models with the evolutionary search for new training data. Theory predicts [23] that the best new training ..."
Abstract
-
Cited by 8 (2 self)
- Add to MetaCart
(Show Context)
In previous work [7] a computational framework was demonstrated that employs evolutionary algorithms to automatically model a given system. This is accomplished by alternating the evolution of models with the evolutionary search for new training data. Theory predicts [23] that the best new training data is that which induces maximum disagreement across the current model set. Here it is demonstrated that in a robot application this is not the case, and alternative fitness functions are developed that seek other, better training data. Also, it is shown that although crossover successfully reduces the mean error of the model set, it compromises the ability of the framework to find new, informative training data. This has implications for how to create adaptive, selfmodeling machines, and suggests how competitive processes in the brain underlie the generation of intelligent behavior.
Modeling human expertise on a cheese ripening industrial process using GP
- in "PPSN, 10th International Conference on Parallel Problem Solving from Nature", September 13-17, Technische Universität
, 2008
"... GP ..."
(Show Context)
Informative Sampling for Large Unbalanced Data Sets
"... Selective sampling is a form of active learning which can reduce the cost of training by only drawing informative data points into the training set. This selected training set is expected to contain more information for modeling compared to random sampling, thus making modeling faster and more accur ..."
Abstract
-
Cited by 4 (3 self)
- Add to MetaCart
(Show Context)
Selective sampling is a form of active learning which can reduce the cost of training by only drawing informative data points into the training set. This selected training set is expected to contain more information for modeling compared to random sampling, thus making modeling faster and more accurate. We introduce a novel approach to selective sampling, which is derived from the Estimation-Exploration Algorithm (EEA). The EEA is a coevolutionary algorithm that uses model disagreement to determine the significance of a training datum, and evolves a set of models only on the selected data. The algorithm in this paper trains a population of Artificial Neural Networks (ANN) on the training set, and uses their disagreement to seek new data for the training set. A medical data set called the National Trauma Data Bank (NTDB) is used to test the algorithm. Experiments show that the algorithm outperforms the equivalent algorithm using randomly-selected data and sampling evenly from each class. Finally, the selected training data reveals which features most affect outcome, allowing for both improved modeling and understanding of the processes that gave rise to the data.
Improving dynamic software analysis by applying grammar inference principles
- Journal of Software Maintenance and Evolution: Research and Practice
, 2008
"... Grammar inference is a family of machine learning techniques that aim to infer grammars from a sample of sentences in some (unknown) language. Dynamic analysis is a family of techniques in the domain of software engineering that attempts to infer rules that govern the behaviour of software systems f ..."
Abstract
-
Cited by 4 (3 self)
- Add to MetaCart
(Show Context)
Grammar inference is a family of machine learning techniques that aim to infer grammars from a sample of sentences in some (unknown) language. Dynamic analysis is a family of techniques in the domain of software engineering that attempts to infer rules that govern the behaviour of software systems from a sample of executions. Despite their disparate domains, both fields have broadly similar aims; they try to infer rules that govern the behaviour of some unknown system from a sample of observations. Deriving general rules about program behaviour from dynamic analysis is difficult because it is virtually impossible to identify and supply a complete sample of necessary program executions. The problems that arise with incomplete input samples have been extensively investigated in the grammar inference community. This has resulted in a number of advances that have produced increasingly sophisticated solutions that are more successful at accurately inferring grammars from (potentially) sparse information about the underlying system. This paper investigates the similarities and shows how many of these advances can be applied with similar effect to dynamic analysis problems by a series of small experiments on random state machines. KEY WORDS: Reverse engineering, dynamic analysis, grammar inference
Grammar-based classifier system: a universal tool for grammatical inference
- WSEAS TRANS. ON COMPUTERS
, 2008
"... Grammatical Inference deals with the problem of learning structural models, such as grammars, from different sort of data patterns, such as artificial languages, natural languages, biosequences, speech and so on. This article describes a new grammatical inference tool, Grammar-based Classifier Syst ..."
Abstract
-
Cited by 3 (0 self)
- Add to MetaCart
(Show Context)
Grammatical Inference deals with the problem of learning structural models, such as grammars, from different sort of data patterns, such as artificial languages, natural languages, biosequences, speech and so on. This article describes a new grammatical inference tool, Grammar-based Classifier System (GCS) dedicated to learn grammar from data. GCS is a new model of Learning Classifier Systems in which the population of classifiers has a form of a context-free grammar rule set in a Chomsky Normal Form. GCS has been proposed to address both regular language induction and the natural language grammar induction as well as learning formal grammar for DNA sequence. In all cases near-optimal solutions or better than reported in the literature were obtained.
Active Learning with Adaptive Heterogeneous Ensembles
"... Abstract—One common approach to active learning is to iteratively train a single classifier by choosing data points based on its uncertainty, but it is nontrivial to design uncertainty measures unbiased by the choice of classifier. Query by committee [1] suggests that given an ensemble of diverse bu ..."
Abstract
-
Cited by 3 (3 self)
- Add to MetaCart
(Show Context)
Abstract—One common approach to active learning is to iteratively train a single classifier by choosing data points based on its uncertainty, but it is nontrivial to design uncertainty measures unbiased by the choice of classifier. Query by committee [1] suggests that given an ensemble of diverse but accurate classifiers, the most informative data points are those that cause maximal disagreement among the predictions of the ensemble members. However the method for finding ensembles appropriate to a given data set remains an open question. In this paper, the random subspace method is combined with active learning to create multiple instances of different classifier types, and an algorithm is introduced that adapts the ratio of different classifier types in the ensemble towards better overall accuracy. Here we show that the proposed algorithm outperforms C4.5 with uncertainty sampling, Naive Bayes with uncertainty sampling, bagging, boosting and the random subspace method with random sampling. To the best of our knowledge, our work is the first to adapt the ratio of classifiers in a heterogeneous ensemble for active learning.
Evaluation and Comparison of Inferred Regular Grammars
- In Proceedings of the International Colloquium on Grammar Inference (ICGI
, 2008
"... Abstract. The accuracy of an inferred grammar is commonly computed by measuring the percentage of sequences that are correctly classified from a random sample of sequences produced by the target grammar. This approach is problematic because (a) it is unlikely that a random sample of sequences will a ..."
Abstract
-
Cited by 3 (2 self)
- Add to MetaCart
(Show Context)
Abstract. The accuracy of an inferred grammar is commonly computed by measuring the percentage of sequences that are correctly classified from a random sample of sequences produced by the target grammar. This approach is problematic because (a) it is unlikely that a random sample of sequences will adequately test the grammar and (b) the use of a single probability value provides little insight into the extent to which a grammar is (in-)accurate. This paper addresses these two problems by proposing the use of established model-based testing techniques from the field of software engineering to systematically generate test sets, along with the use of the Precision and Recall measure from the field of information retrieval to concisely represent the accuracy of the inferred machine.
Adaptive Informative Sampling for Active Learning
, 2009
"... Many approaches to active learning involve periodically training one classifier and choosing data points with the lowest confidence, but designing a confidence measure is nontrivial. An alternative approach is to periodically choose data instances that maximize disagreement among the label predictio ..."
Abstract
-
Cited by 2 (0 self)
- Add to MetaCart
(Show Context)
Many approaches to active learning involve periodically training one classifier and choosing data points with the lowest confidence, but designing a confidence measure is nontrivial. An alternative approach is to periodically choose data instances that maximize disagreement among the label predictions across an ensemble of classifiers. Many classifiers with different underlying structures could fit this framework, but some ensembles are more suitable for some data sets than others. The question then arises as to how to find the most suitable ensemble for a given data set. In this work we introduce a method that begins with a heterogeneous ensemble composed of multiple instances of different classifier types, which we call adaptive informative sampling. The algorithm periodically adds data points to the training set, adapts the ratio of classifier types in the heterogeneous ensemble in favor of the better classifier type, and optimizes the classifiers in the ensemble using stochastic methods. Experimental results show that the proposed method performs consistently better than homogeneous ensembles. Comparison with random sampling and uncertainty sampling shows that the algorithm effectively draws informative data points for training. 1
Combined Structure and Motion Extraction from Visual Data Using Evolutionary Active Learning
"... We present a novel stereo vision modeling framework that generates approximate, yet physically-plausible representations of objects rather than creating accurate models that are computationally expensive to generate. Our approach to the modeling of target scenes is based on carefully selecting a sma ..."
Abstract
-
Cited by 2 (0 self)
- Add to MetaCart
(Show Context)
We present a novel stereo vision modeling framework that generates approximate, yet physically-plausible representations of objects rather than creating accurate models that are computationally expensive to generate. Our approach to the modeling of target scenes is based on carefully selecting a small subset of the total pixels available for visual processing. To achieve this, we use the estimation-exploration algorithm (EEA) to create the visual models: a population of three-dimensional models is optimized against a growing set of training pixels, and periodically a new pixel that causes disagreement among the models is selected from the observed stereo images of the scene and added to the training set. We show here that using only 5 % of the available pixels, the algorithm can generate approximate models of compound objects in a scene. Our algorithm serves the dual goals of extracting the 3D structure and relative motion of objects of interest by modeling the target objects in terms of their physical parameters (e.g., position, orientation, shape, etc.), and tracking how these parameters vary with time. We support our claims with results from simulation as well from a real robot lifting a compound object.