Heterogeneous Ensemble for Imaginary Scene Classification

Artificial Intelligence XXXIV

2017

Self Cite

Abstract. Multimedia data consists of several different types of data, such as numbers, text, images, audio etc. and they usually need to be fused or integrated before analysis. This study investigates a feature-level aggregation approach to combine multimedia datasets for building heterogeneous ensembles for classification. It firstly aggregates multimedia datasets at feature level to form a normalised big dataset, then uses some parts of it to generate classifiers with different learning algorithms. Finally it applies three rules to select appropriate classifiers based on their accuracy and/or diversity to build heterogeneous ensembles. The method is tested on a multimedia dataset and the results show that the heterogeneous ensembles outperform the individual classifiers as well as homogeneous ensembles. However, it should be noted that, it is possible on some cases that the combined dataset does not produce better results than using single media data.

Section: Experiments Design and Resultsmentioning

confidence: 99%

“…2. Main steps for R0, R1 and R2 in HES [14] from the sorted PM are selected (Equation 2) and added to the ensemble, Φ. Therefore Φ now contains MAM and the (N-1) most diverse models from PM.…”

Section: Rules For Selecting Modelsmentioning

confidence: 99%

Feature Level Ensemble Method for Classifying Multi-media Data

Alyahyan

Artificial Intelligence XXXIV

2017

Self Cite

“…Rules R0, R1 and R2 are described in our earlier work [11] [12]. We give a brief summary of them here for convenience as they are the bases of the new rule.…”

Section: The Decision Level Ensemble Methods Frameworkmentioning

confidence: 99%

“…The results of GDLEM were compared with the feature-level ensemble method(FLEM) and various heterogeneous ensembles based on the single media data, text (HEST) and image data (HESG). The full comparative results between the FLEM and the HESG were published in [12] and the full results for the HEST were published in [11]. Figure.…”

Section: Critical Comparison With Other Ensemblesmentioning

confidence: 99%

See 1 more Smart Citation

Generalised Decision Level Ensemble Method for Classifying Multi-media Data

Alyahyan

Lecture Notes in Computer Science

2018

Self Cite

In recent decades, multimedia data have been commonly generated and used in various domains, such as in healthcare and social media due to their ability of capturing rich information. But as they are unstructured and separated, how to fuse and integrate multimedia datasets and then learn from them eectively have been a main challenge to machine learning. We present a novel generalised decision level ensemble method (GDLEM) that combines the multimedia datasets at decision level. After extracting features from each of multimedia datasets separately, the method trains models independently on each media dataset and then employs a generalised selection function to choose the appropriate models to construct a heterogeneous ensemble. The selection function is dened as a weighted combination of two criteria: the accuracy of individual models and the diversity among the models. The framework is tested on multimedia data and compared with other heterogeneous ensembles. The results show that the GDLEM is more exible and eective.

Weighted Ensemble Methods for Predicting Train Delays

Ghamdi

Parr

Computational Science and Its Applications – ICCSA 2020

2020

Self Cite

Train delays have become a serious and common problem in the rail services due to the increasing number of passengers and limited rail network capacity, so being able to predict train delays accurately is essential for train controllers to devise appropriate plans to prevent or reduce some delays. This paper presents a machine learning ensemble framework to improve the accuracy and consistency of train delay prediction. The basic idea is to train many different types of machine learning models for each station along a chosen journey of train service using historical data and relevant weather data, and then with certain criteria to choose some models to build an ensemble. It then combines the outputs from its member models with an aggregation function to produce the final prediction. Two aggregation functions were devised to combine the outputs of individual models: averaging and weighted averaging. These ensembles were implemented with a framework and their performance was tested with the data from an intercity train service as a case study. The accuracy was measured by the percentages of correct prediction of the arrival time for a train and correct prediction within one minute to the actual arrival time. The mean accuracies and standard deviations are 42.3%( ) from the individual models, 57.8%( ) from the averaging ensembles, and 72.8%( ) from the weighted ensembles. For the predictions within one minute of the actual times, they are 86.4%( ), 94.6%( ) and 96.0%( ) respectively. So overall, the ensembles significantly improved not only the prediction accuracies but also the consistency and the weighted ensembles are clearly the best.