scholarly journals A logical framework to study concept-learning biases in the presence of multiple explanations

2021 ◽  
Author(s):  
Sergio Abriola ◽  
Pablo Tano ◽  
Sergio Romano ◽  
Santiago Figueira
Keyword(s):  
Eye Tracking ◽  
Concept Learning ◽  
Propositional Logic ◽  
Classification Rule ◽  
Classification Rules ◽  
Logical Framework ◽  
Transfer Effects ◽  
Learning Biases

AbstractWhen people seek to understand concepts from an incomplete set of examples and counterexamples, there is usually an exponentially large number of classification rules that can correctly classify the observed data, depending on which features of the examples are used to construct these rules. A mechanistic approximation of human concept-learning should help to explain how humans prefer some rules over others when there are many that can be used to correctly classify the observed data. Here, we exploit the tools of propositional logic to develop an experimental framework that controls the minimal rules that are simultaneously consistent with the presented examples. For example, our framework allows us to present participants with concepts consistent with a disjunction and also with a conjunction, depending on which features are used to build the rule. Similarly, it allows us to present concepts that are simultaneously consistent with two or more rules of different complexity and using different features. Importantly, our framework fully controls which minimal rules compete to explain the examples and is able to recover the features used by the participant to build the classification rule, without relying on supplementary attention-tracking mechanisms (e.g. eye-tracking). We exploit our framework in an experiment with a sequence of such competitive trials, illustrating the emergence of various transfer effects that bias participants’ prior attention to specific sets of features during learning.

Ant Miner

2020 ◽  
Vol 10 (1) ◽  
pp. 45-59
Author(s):  
Bijaya Kumar Nanda ◽  
Satchidananda Dehuri
Keyword(s):  
Data Mining ◽  
Large Data ◽  
Classification Rule ◽  
Classification Rules ◽  
Rule Mining ◽  
Ant Colonies ◽  
Benchmark Datasets ◽  
Objective Classification ◽  
Single Objective ◽  
Better Than

In data mining the task of extracting classification rules from large data is an important task and is gaining considerable attention. This article presents a novel ant miner for classification rule mining. The ant miner is inspired by researches on the behaviour of real ant colonies, simulated annealing, and some data mining concepts as well as principles. This paper presents a Pittsburgh style approach for single objective classification rule mining. The algorithm is tested on a few benchmark datasets drawn from UCI repository. The experimental outcomes confirm that ant miner-HPB (Hybrid Pittsburgh Style Classification) is significantly better than ant-miner-PB (Pittsburgh Style Classification).

scholarly journals A New Classification Approach Based on Multiple Classification Rules

2014 ◽  
Vol 2014 ◽  
pp. 1-7 ◽  
Author(s):  
Zhongmei Zhou
Keyword(s):  
Classification Rule ◽  
Classification Rules ◽  
Associative Classification ◽  
New Classification ◽  
Class Label ◽  
Rule Based ◽  
Classification Approach ◽  
Minimum Support ◽  
Multiple Classification ◽  
Rule Set

A good classifier can correctly predict new data for which the class label is unknown, so it is important to construct a high accuracy classifier. Hence, classification techniques are much useful in ubiquitous computing. Associative classification achieves higher classification accuracy than some traditional rule-based classification approaches. However, the approach also has two major deficiencies. First, it generates a very large number of association classification rules, especially when the minimum support is set to be low. It is difficult to select a high quality rule set for classification. Second, the accuracy of associative classification depends on the setting of the minimum support and the minimum confidence. In comparison with associative classification, some improved traditional rule-based classification approaches often produce a classification rule set that plays an important role in prediction. Thus, some improved traditional rule-based classification approaches not only achieve better efficiency than associative classification but also get higher accuracy. In this paper, we put forward a new classification approach called CMR (classification based on multiple classification rules). CMR combines the advantages of both associative classification and rule-based classification. Our experimental results show that CMR gets higher accuracy than some traditional rule-based classification methods.

scholarly journals Can SLE classification rules be effectively applied to diagnose unclear SLE cases?

Lupus ◽  
2016 ◽  
Vol 26 (2) ◽  
pp. 150-162 ◽  
Author(s):  
A Mesa ◽  
M Fernandez ◽  
W Wu ◽  
G Narasimhan ◽  
EL Greidinger ◽  
...  
Keyword(s):  
Lupus Erythematosus ◽  
Operating Characteristic ◽  
Clinical Manifestations ◽  
Classification Criteria ◽  
Classification Rule ◽  
Classification Rules ◽  
Clinical Markers ◽  
Systemic Lupus ◽  
Reduced Models ◽  
Logistic Regressions

Objective The objective of this paper is to develop novel classification criteria to distinguish between unclear systemic lupus erythematosus (SLE) and mixed connective tissue disease (MCTD) cases. Methods A total of 205 variables from 111 SLE and 55 MCTD patients were evaluated to uncover unique molecular and clinical markers for each disease. Binomial logistic regressions (BLRs) were performed on currently used SLE and MCTD classification criteria sets to obtain six reduced models with power to discriminate between unclear SLE and MCTD patients that were confirmed by receiving operating characteristic (ROC) curve. Decision trees were employed to delineate novel classification rules to discriminate between unclear SLE and MCTD patients. Results SLE and MCTD patients exhibited contrasting molecular markers and clinical manifestations. Furthermore, reduced models highlighted SLE patients exhibiting prevalence of skin rashes and renal disease while MCTD cases show dominance of myositis and muscle weakness. Additionally decision tree analyses revealed a novel classification rule tailored to differentiate unclear SLE and MCTD patients (Lu-vs-M) with an overall accuracy of 88%. Conclusions Validation of our novel proposed classification rule (Lu-vs-M) includes novel contrasting characteristics (calcinosis, CPK elevated and anti-IgM reactivity for U1-70K, U1A and U1C) between SLE and MCTD patients and showed a 33% improvement in distinguishing these disorders when compared to currently used classification criteria sets. Pending additional validation, our novel classification rule is a promising method to distinguish between patients with unclear SLE and MCTD diagnosis.

SKEPTICAL ABDUCTION

1993 ◽  
Vol 02 (04) ◽  
pp. 511-540 ◽  
Author(s):  
P. MARQUIS
Keyword(s):  
Propositional Logic ◽  
Order Logic ◽  
First Order Logic ◽  
Real Problem ◽  
Logical Framework ◽  
First Order ◽  
Selective Generation ◽  
The Given ◽  
Reasonable Prospect

Abduction is the process of generating the best explanation as to why a fact is observed given what is already known. A real problem in this area is the selective generation of hypotheses that have some reasonable prospect of being valid. In this paper, we propose the notion of skeptical abduction as a model to face this problem. Intuitively, the hypotheses pointed out by skeptical abduction are all the explanations that are consistent with the given knowledge and that are minimal, i.e. not unnecessarily general. Our contribution is twofold. First, we present a formal characterization of skeptical abduction in a logical framework. On this ground, we address the problem of mechanizing skeptical abduction. A new method to compute minimal and consistent hypotheses in propositional logic is put forward. The extent to which skeptical abduction can be mechanized in first—order logic is also investigated. In particular, two classes of first-order formulas in which skeptical abduction is effective are provided. As an illustration, we finally sketch how our notion of skeptical abduction applies as a theoretical tool to some artificial intelligence problems (e.g. diagnosis, machine learning).

Self-Adaptive Weighting Text Association Categorization Algorithm Research

2010 ◽  
Vol 171-172 ◽  
pp. 246-251
Author(s):  
Liang Jun Li ◽  
Bin Zhang ◽  
Yuan Yuan Che ◽  
Ming Yang ◽  
Tie Nan Li
Keyword(s):  
Training Sample ◽  
Classification Rule ◽  
Classification Algorithms ◽  
Classification Rules ◽  
Sample Collection ◽  
Adaptive Weighting ◽  
Adjustment Factors ◽  
Weighing Method ◽  
The Stability ◽  
Self Adaptive

In text association classification research, feature distribution of the training sample collection impacts greatly on the classification results, even with a same classification algorithm classification results will have obvious differences using different sample collections. In order to solve the problem, the stability of association classification is improved by the weighing method in the paper, the design realizes the association classification algorithms (WARC) based on rule weight. In the WARC algorithm, this paper proposes the concept of classification rule intensity and gives the concrete formula. Using rule intensity defines the rule adjustment factors that adjust uneven classification rules. Experimental results show the accuracy of text classification can be improved obviously by self-adaptive weighting.

scholarly journals Inferring Agents’ Goals from Observing Successful Traces

2021 ◽  
Vol 11 (9) ◽  
pp. 4116
Author(s):  
Guillaume Lorthioir ◽  
Katsumi Inoue ◽  
Gauvain Bourgne
Keyword(s):  
Concept Learning ◽  
Propositional Logic ◽  
Learning Algorithm ◽  
Search Space ◽  
Plan Recognition ◽  
Goal Recognition

Goal recognition is a sub-field of plan recognition that focuses on the goals of an agent. Current approaches in goal recognition have not yet tried to apply concept learning to a propositional logic formalism. In this paper, we extend our method for inferring an agent’s possible goal by observing this agent in a series of successful attempts to reach its goal and using concept learning on these observations. We propose an algorithm, LFST (Learning From Successful Traces), to produce concise hypotheses about the agent’s goal. We show that if such a goal exists, our algorithm always provides a possible goal for the agent, and we evaluate the performance of our algorithm in different settings. We compare it to another concept-learning algorithm that uses a formalism close to ours, and we obtain better results at producing the hypotheses with our algorithm. We introduce a way to use assumptions about the agent’s behavior and the dynamics of the environment, thus improving the agent’s goal deduction by optimizing the potential goals’ search space.

A Fast Boosting Based Incremental Genetic Algorithm for Mining Classification Rules in Large Datasets

2011 ◽  
Vol 2 (1) ◽  
pp. 49-58
Author(s):  
Periasamy Vivekanandan ◽  
Raju Nedunchezhian
Keyword(s):  
Genetic Algorithm ◽  
Training Data ◽  
Classification Rule ◽  
Rule Discovery ◽  
Classification Rules ◽  
Search Technique ◽  
Natural Evolution ◽  
Ensemble Of Classifiers ◽  
Data Set ◽  
Mining Community

Genetic algorithm is a search technique purely based on natural evolution process. It is widely used by the data mining community for classification rule discovery in complex domains. During the learning process it makes several passes over the data set for determining the accuracy of the potential rules. Due to this characteristic it becomes an extremely I/O intensive slow process. It is particularly difficult to apply GA when the training data set becomes too large and not fully available. An incremental Genetic algorithm based on boosting phenomenon is proposed in this paper which constructs a weak ensemble of classifiers in a fast incremental manner and thus tries to reduce the learning cost considerably.

CoABCMiner: An Algorithm for Cooperative Rule Classification System Based on Artificial Bee Colony

2016 ◽  
Vol 25 (01) ◽  
pp. 1550028 ◽  
Author(s):  
Mete Celik ◽  
Fehim Koylu ◽  
Dervis Karaboga
Keyword(s):  
Artificial Bee Colony ◽  
Statistical Tests ◽  
Rule Learning ◽  
Machine Learning Algorithms ◽  
Classification Model ◽  
Classification Rule ◽  
Data Sets ◽  
Classification Rules ◽  
Data Set ◽  
Bee Colony

In data mining, classification rule learning extracts the knowledge in the representation of IF_THEN rule which is comprehensive and readable. It is a challenging problem due to the complexity of data sets. Various meta-heuristic machine learning algorithms are proposed for rule learning. Cooperative rule learning is the discovery process of all classification rules with a single run concurrently. In this paper, a novel cooperative rule learning algorithm, called CoABCMiner, based on Artificial Bee Colony is introduced. The proposed algorithm handles the training data set and discovers the classification model containing the rule list. Token competition, new updating strategy used in onlooker and employed phases, and new scout bee mechanism are proposed in CoABCMiner to achieve cooperative learning of different rules belonging to different classes. We compared the results of CoABCMiner with several state-of-the-art algorithms using 14 benchmark data sets. Non parametric statistical tests, such as Friedman test, post hoc test, and contrast estimation based on medians are performed. Nonparametric tests determine the similarity of control algorithm among other algorithms on multiple problems. Sensitivity analysis of CoABCMiner is conducted. It is concluded that CoABCMiner can be used to discover classification rules for the data sets used in experiments, efficiently.

scholarly journals Classification Rule Extraction Based on Relevant, Irredundant Attributes and Rule Enlargement

2007 ◽  
Vol 11 (4) ◽  
pp. 389-395
Author(s):  
George Lashkia ◽  
◽  
Laurence Anthony ◽  
Hiroyasu Koshimizu ◽  
Keyword(s):  
Extraction Method ◽  
Irrelevant Information ◽  
Rule Extraction ◽  
Classification Rule ◽  
Classification Rules ◽  
Simple Rules ◽  
Conventional Methods

In this paper we focus on the induction of classification rules from examples. Conventional algorithms fail in discovering effective knowledge when the database contains irrelevant information. We present a new rule extraction method, RGT, which tackles this problem by employing only relevant and irredundant attributes. Simplicity of rules is also our major concern. In order to create simple rules, we estimate the purity of patterns and propose a rule enlargement approach, which consists of rule merging and rule expanding procedures. In this paper, we describe the methodology for the RGT algorithm, discuss its properties, and compare it with conventional methods.

scholarly journals Application of spatial classification rules for remotely sensed images

2014 ◽  
Vol 55 ◽  
Author(s):  
Giedrius Stabingis ◽  
Lijana Stabingienė
Keyword(s):  
Classification Accuracy ◽  
Supervised Classification ◽  
Spatial Information ◽  
Gaussian Random Field ◽  
Remotely Sensed ◽  
Classification Rule ◽  
Classification Rules ◽  
Classification Methods ◽  
Remotely Sensed Images ◽  
Supervised Classification Methods

In this paper the remote sensed image classification example using spacial classification rule with distance (SCRD) is examined. This supervised classification method was first presented in paper [11]. This method is improved version of earlier method PBDF [4, 10, 9], during the classification it incorporates more spatial information. The advantage of this method is its ability to classify data which is corrupted by Gaussian random field and it is typical to remotely sensed images classified in this letter which are corrupted by clouds. Classification accuracy is compared with earlier method and with other commonly used supervised classification methods.

Sign in / Sign up

Export Citation Format

Share Document