Effectiveness Of Machine Learning Algorithms Research Articles

Evaluating the Effectiveness of Machine Learning Algorithms in Predictive Modelling

Predictive modelling is a statistical technique to predict future behaviour. Machine learning is one of the most popular methods for predicting the future behaviour. From the plethora of algorithms available it is always interesting to find out which algorithm or technique is most suitable for data under consideration. Educational Data Mining is the area of research where predictive modelling is most useful. Predicting the grades of the undergraduate students accurately can help students as well as educators in many ways. Early prediction can help motivating students in better ways to select their future endeavour. This paper presents the results of various machine learning algorithms applied to the data collected from undergraduate studies. It evaluates the effectiveness of various machine learning algorithms when applied to data collected from undergraduate studies. Two major challenges are addressed as: choosing the right features and choosing the right algorithm for prediction.

Machine learning techniques in searches for tt̄h in the h → bb̄ decay channel

Study of the production of pairs of top quarks in association with a Higgs boson is one of the primary goals of the Large Hadron Collider over the next decade, as measurements of this process may help us to understand whether the uniquely large mass of the top quark plays a special role in electroweak symmetry breaking. Higgs bosons decay predominantly to bb̄, yielding signatures for the signal that are similar to tt̄ + jets with heavy flavor. Though particularly challenging to study due to the similar kinematics between signal and background events, such final states (tt̄ bb̄) are an important channel for studying the top quark Yukawa coupling. This paper presents a systematic study of machine learning (ML) methods for detecting tt̄h in the h → bb̄ decay channel. Among the eight ML methods tested, we show that two models, extreme gradient boosted trees and neural network models, outperform alternative methods. We further study the effectiveness of ML algorithms by investigating the impact of feature set and data size, as well as the structure of the models. While extended feature set and larger training sets expectedly lead to improvement of performance, shallow models deliver comparable or better performance than their deeper counterparts. Our study suggests that ensembles of trees and neurons, not necessarily deep, work effectively for the problem of tt̄h detection.

AUTOMATED DIAGNOSIS IN A POPULATION-BASED SCREENING FOR LUNG CANCER

Oncologists nowadays are faced with big amount of heterogeneous medical data of diagnostic studies. Possible errors in determining the nature and extent of spread the tumor process will inevitably reduce the effectiveness of treatment and increase the unnecessary costs to it. To reduce the burden on clinicians, various computer-aided solutions based on machine learning algorithms are being developed. We made an attempt to evaluate effectiveness of thirteen machine learning algorithms in the tasks of classification of pathologic tissue samples in cancerous thorax based on gene expression levels. For a preliminary study we used open data set of molecular genetics composition of lung adenocarcinoma and pleural mesothelioma. Effectiveness of machine learning algorithms was evaluated by Matthews correlation coefficient and Area Under ROC Curve. Best results were showed by two methods: Bayesian logistic regression and Discriminative Multinomial Naive Bayes classifier. Nevertheless, all of the methods were effective at automatic discrimination of two types of cancer. That proves machine learning algorithms are applicable in lung cancer classification. In the future studies it will be carried out a similar analysis of the diagnostic value of methods for other malignancies with more complex differential morphological diagnosis. Similar methods can be applied to other diagnostic studies including computerized tomography image analysis in the differential diagnosis of lung nodules.