Instance-based Approach Research Articles

An instance theory of associative learning

We present and test an instance model of associative learning. The model, Minerva-AL, treats associative learning as cued recall. Memory preserves the events of individual trials in separate traces. A probe presented to memory contacts all traces in parallel and retrieves a weighted sum of the traces, a structure called the echo. Learning of a cue-outcome relationship is measured by the cue's ability to retrieve a target outcome. The theory predicts a number of associative learning phenomena, including acquisition, extinction, reacquisition, conditioned inhibition, external inhibition, latent inhibition, discrimination, generalization, blocking, overshadowing, overexpectation, superconditioning, recovery from blocking, recovery from overshadowing, recovery from overexpectation, backward blocking, backward conditioned inhibition, and second-order retrospective revaluation. We argue that associative learning is consistent with an instance-based approach to learning and memory.

Multi-model strategy based evidential soft sensor model for predicting evaluation of variables with uncertainty

The traditional soft sensor methods cannot be applied in the cases where uncertainty exists. To predict evaluation of variable with uncertainty, a novel method, called multi-evidential soft sensor model, was proposed based on evidence theory and multi-model strategy. In such method, the overall multi-model on the whole data space is established by weight sum of the multiple local models on local data spaces. For arbitrary given input vector x, each local model provides a prediction regarding the value of the output variable y, in the form of fuzzy belief assignment (FBA), defined as a collection of fuzzy sets of values with associated masses of belief. The output FBA is computed using a parametric, instance-based approach: imprecise training samples in the neighbourhood of x are considered as sources of partial information on the response variable y, and provide pieces of evidence reflecting the values taken by the response y; the pieces of evidences are pooled by using Dempster's rule of combination. To identify the parameters involved in the overall and local models, the so-called imprecise training samples are constructed from running data by means of data analysis and expertise knowledge.To validate the proposed method, a numerical experiment was designed based on a UCI dataset, and the experimental results suggest its power for predicting evaluation of variables with uncertainty. Finally, the prediction of unmeasured parameter level of coal powder filling in tubular ball mill was taken as the engineering example to validate the proposed method. The predicting results are in line with the experts’ analysis. The two validations suggest that the proposed method is applicable for predicting variables with uncertainty in process industry.

Log integration on large scale for global networking monitoring

Supposing that the overall situation is dug out from the distributed monitoring nodes, there should be two critical obstacles, heterogenous schema and instance, to integrating heterogeneous data from different monitoring sensors. To tackle the challenge of heterogenous schema, an instance-based approach for schema mapping, named instance-based machine-learning (IML) approach was described. And to solve the problem of heterogenous instance, a novel approach, called statistic-based clustering (SBC) approach, which utilized clustering and statistics technologies to match large scale sources holistically, was also proposed. These two algorithms utilized the machine-leaning and clustering technology to improve the accuracy. Experimental analysis shows that the IML approach is more precise than SBC approach, reaching at least precision of 81% and recall rate of 82%. Simulation studies further show that SBC can tackle large scale sources holistically with 85% recall rate when there are 38 data sources.

Reasoning about XML update constraints

We introduce in this paper a class of constraints for describing how an XML document can evolve, namely XML update constraints. For these constraints, we study the implication problem, giving algorithms and complexity results for constraints of varying expressive power. Besides classical constraint implication, we also consider an instance-based approach in which we take into account data. More precisely, we study implication with respect to a current tree instance, resulting from a series of unknown updates. The main motivation of our work is reasoning about data integrity under update restrictions in contexts where owners may lose control over their data, such as in publishing or exchange.

Data and Knowledge Driven Adaptive Evidential Soft Sensor Model for Monitoring Unmeasured Parameters

AbstractTo solve the monitoring or predicting for the unmeasured parameters, which can not be monitored online and offline by device in industrial processes, a method, called data and knowledge driven adaptive evidential soft sensor model, was proposed based on evidence theory. In the proposed method, the imprecise training samples, applied to identify the proposed soft sensor, are constructed from running data by means of data analysis and expertise knowledge. For a given input vector x, the method provides a prediction regarding the value of the output variable y, in the form of fuzzy belief assignment (FBA), defined as a collection of fuzzy sets of values with associated masses of belief. The output FBA is computed using a nonparametric, instance-based approach: imprecise training samples in the neighborhood of x are considered as sources of partial information on the response variable; the pieces of evidence are discounted as a function of their distance to x, and pooled by using Dempster's rule of combination. Taking into account the unmeasured parameter level of coal powder filling in tubular ball mill as the example, some experiments were conducted to validate the proposed method.

Open-ended category learning for language acquisition

Motivated by the need to support language-based communication between robots and their human users, as well as grounded symbolic reasoning, this paper presents a learning architecture that can be used by robotic agents for long-term and open-ended category acquisition. To be more adaptive and to improve learning performance as well as memory usage, this learning architecture includes a metacognitive processing component. Multiple object representations and multiple classifiers and classifier combinations are used. At the object level, the main similarity measure is based on a multi-resolution matching algorithm. Categories are represented as sets of known instances. In this instance-based approach, storing and forgetting rules optimise memory usage. Classifier combinations are based on majority voting and the Dempster–Shafer evidence theory. All learning computations are carried out during the normal execution of the agent, which allows continuous monitoring of the performance of the different classifiers. The measured classification successes of the individual classifiers support an attentional selection mechanism, through which classifier combinations are dynamically reconfigured and a specific classifier is chosen to predict the category of a new unseen object. A simple physical agent, incorporating these learning capabilities, is used to test the approach. A long-term experiment was carried out having in mind the open-ended nature of category learning. With the help of a human mediator, the agent incrementally learned 68 categories of real-world objects visually perceivable through an inexpensive camera. Various aspects of the approach are evaluated through systematic experiments.

An Instance-Based Method for Remaining Useful Life Estimation for Aircraft Engines

Under customer service agreements (CSA), engine operational data are collected and stored for monitoring and analysis. Other data sources provide damage assessments that are either provided post-maintenance or analytically assessed. This paper takes advantage of these data and investigates local fuzzy models to determine the remaining useful life (RUL) of an engine or engine component. Local fuzzy models are related to both kernel regressions and locally weighted learning. The particular local models described in this paper are not based on individual models that consider the track history of a specific engine nor are they based on a global average model that would consider the collective track history of all the engines. Instead, for a given engine or component, this local fuzzy model defines a cluster of peers in which each of these peers is a similar instance to this given engine with comparable operational characteristics; the RUL prediction for this given engine is obtained by a fuzzy aggregation of its peers’ RUL. We combine the fuzzy instance-based approach with an evolutionary framework for model tuning and maintenance. This evolutionary tuning process is repeated periodically to automatically update and improve the fuzzy models such that they can be updated to date with the latest collection of data. This fuzzy instance-based approach is applied to predicting the RUL of a commercial engine validated with post-maintenance assessment.

Extracting Web Data Using Instance-Based Learning

This paper studies structured data extraction from Web pages. Existing approaches to data extraction include wrapper induction and automated methods. In this paper, we propose an instance-based learning method, which performs extraction by comparing each new instance to be extracted with labeled instances. The key advantage of our method is that it does not require an initial set of labeled pages to learn extraction rules as in wrapper induction. Instead, the algorithm is able to start extraction from a single labeled instance. Only when a new instance cannot be extracted does it need labeling. This avoids unnecessary page labeling, which solves a major problem with inductive learning (or wrapper induction), i.e., the set of labeled instances may not be representative of all other instances. The instance-based approach is very natural because structured data on the Web usually follow some fixed templates. Pages of the same template usually can be extracted based on a single page instance of the template. A novel technique is proposed to match a new instance with a manually labeled instance and in the process to extract the required data items from the new instance. The technique is also very efficient. Experimental results based on 1,200 pages from 24 diverse Web sites demonstrate the effectiveness of the method. It also outperforms the state-of-the-art existing systems significantly.

An instance-based learning approach based on grey relational structure

In instance-based learning, the `nearness' between two instances--used for pattern classification--is generally determined by some similarity functions, such as the Euclidean or Value Difference Metric (VDM). However, Euclidean-like similarity functions are normally only suitable for domains with numeric attributes. The VDM metrics are mainly applicable to domains with symbolic attributes, and their complexity increases with the number of classes in a specific application domain. This paper proposes an instance-based learning approach to alleviate these shortcomings. Grey relational analysis is used to precisely describe the entire relational structure of all instances in a specific domain. By using the grey relational structure, new instances can be classified with high accuracy. Moreover, the total number of classes in a specific domain does not affect the complexity of the proposed approach. Forty classification problems are used for performance comparison. Experimental results show that the proposed approach yields higher performance over other methods that adopt one of the above similarity functions or both. Meanwhile, the proposed method can yield higher performance, compared to some other classification algorithms.

Instance-Based Regression by Partitioning Feature Projections

A new instance-based learning method is presented for regression problems with high-dimensional data. As an instance-based approach, the conventional method, KNN, is very popular for classification. Although KNN performs well on classification tasks, it does not perform as well on regression problems. We have developed a new instance-based method, called Regression by Partitioning Feature Projections (RPFP) which is designed to meet the requirement for a lazy method that achieves high levels of accuracy on regression problems. RPFP gives better performance than well-known eager approaches found in machine learning and statistics such as MARS, rule-based regression, and regression tree induction systems. The most important property of RPFP is that it is a projection-based approach that can handle interactions. We show that it outperforms existing eager or lazy approaches on many domains when there are many missing values in the training data.

Nonparametric regression analysis of uncertain and imprecise data using belief functions

This paper introduces a new approach to regression analysis based on a fuzzy extension of belief function theory. For a given input vector x , the method provides a prediction regarding the value of the output variable y, in the form of a fuzzy belief assignment (FBA), defined as a collection of fuzzy sets of values with associated masses of belief. The output FBA is computed using a nonparametric, instance-based approach: training samples in the neighborhood of x are considered as sources of partial information on the response variable; the pieces of evidence are discounted as a function of their distance to x , and pooled using Dempster’s rule of combination. The method can cope with heterogeneous training data, including numbers, intervals, fuzzy numbers, and, more generally, fuzzy belief assignments, a convenient formalism for modeling unreliable and imprecise information provided by experts or multi-sensor systems. The performances of the method are compared to those of standard regression techniques using several simulated data sets.

INPUT-DRIVEN LANGUAGE LEARNING

Input-driven models provide an explicit and readily testable account of language learning. Although we share Ellis's view that the statistical structure of the linguistic environment is a crucial and, until recently, relatively neglected variable in language learning, we also recognize that the approach makes three assumptions about cognition and language learning that are not universally shared. The three assumptions concern (a) the language learner as an intuitive statistician, (b) the constraints on what constitute relevant surface cues, and (c) the redescription problem faced by any system that seeks to derive abstract grammatical relations from the frequency of co-occurring surface forms and functions. These are significant assumptions that must be established if input-driven models are to gain wider acceptance. We comment on these issues and briefly describe a distributed, instance-based approach that retains the key features of the input-driven account advocated by Ellis but that also addresses shortcomings of the current approaches.

An instance-based approach to pattern association learning with application to the English past tense verb domain

We present a method for using instance-based learning to acquire pattern association rules and apply it to the English past tense verb domain. To retrieve exemplars, we introduce a distance metric for pattern association, PAMELA, which extends that used in PEBLS for classification. The associated pattern is then built by adapting that of the retrieved exemplar(s). We show that our algorithm IBPA-3, which uses exemplar distance weighting and attribute weighting, improves upon the C4.5-based SPA algorithm and, when tested on the difficult case of irregular verbs, out-performs the current state of the art algorithm for this problem, the relational learner, FOIDL.

Discussion

Lewis (1999) argued that effects of age of acquisition (AoA) are entirely attributable to cumulative frequency. He reported an instance-based model in which the number of instances of the stimulus stored in memory predicts reaction time. We note four aspects of the literature on AoA that cannot be explained by this instance-based approach. Firstly, an effect of AoA has been observed in the absence of an effect of frequency. Secondly, an effect of AoA has been observed when cumulative frequency has been controlled. Thirdly, the effect of AoA is dependent on task. Fourthly, the effect of word frequency is dependent on stimulus modality. Lewis reported an experiment in which participants make a decision based on identity-specific semantic information to celebrity faces to demonstrate an effect of the number of instances in memory, which he interpreted as an effect of AoA. We note that effects of AoA have been found in lexical and perceptual tasks, but to date all attempts to demonstrate an advantage for early-acquired items in semantic classification tasks have failed. We conclude that the effects of AoA cannot be attributed solely to the effects of cumulative frequency.

LearnIT: An Instance-Based Approach to Learning and Reusing Design Strategies

We describe LearnIT, a computer program that can observe an iterative solution to a parametric design problem and learn the design strategy employed. When the design requirements change, the program uses the learned strategy to automatically generate a new solution in the “style” of the original. The program uses a specialized instance-based learning method based on the observation that iterative design is often a form of debugging—each iteration is an attempt to repair a particular flaw in the design. Thus, the program learns the design strategy by observing what actions are taken in response to each kind of flaw. [S1050-0472(00)01203-4]