Performance of Machine Learning Algorithms and Diversity in Data

Abstract

Recent world events in go games between human and artificial intelligence called AlphaGo showed the big advancement in machine learning technologies. While AlphaGo was trained using real world data, AlphaGo Zero was trained using massive random data, and the fact that AlphaGo Zero won AlphaGo completely revealed that diversity and size in training data is important for better performance for the machine learning algorithms, especially in deep learning algorithms of neural networks. On the other hand, artificial neural networks and decision trees are widely accepted machine learning algorithms because of their robustness in errors and comprehensibility respectively. In this paper in order to prove that diversity and size in data are important factors for better performance of machine learning algorithms empirically, the two representative algorithms are used for experiment. A real world data set called breast tissue was chosen, because the data set consists of real numbers that is very good property for artificial random data generation. The result of the experiment proved the fact that the diversity and size of data are very important factors for better performance.

Remote Control of Greenhouse Vegetable Production with Artificial Intelligence—Greenhouse Climate, Irrigation, and Crop Production

“…The predictive and prescriptive power of the algorithms depended partly on the training datasets, and may have restricted algorithms in making early-season crop management decisions. Algorithms capture information present in the data, and the limited variation among the attributes in the training datasets may have affected the performance of the algorithms in the beginning of the experimental period [65]. However, the idea of the experiment was that over the course of the season, the AI algorithms would be improved on the basis of provided management decisions and their effects on greenhouse climate and crop production.…”

How effective is machine learning in stock market predictions?

“…However, what constitutes a reasonable data size is uncertain and context-dependent. Typically, decision trees, naïve Bayes and support vector machine algorithms tend to perform better on small and medium-sized datasets, while random forests, k-nearest neighbors, and artificial neural networks may perform better on larger datasets [ 38 , 39 ]. Similar studies in the literature have been observed to use datasets of different sizes.…”

DyViR: dynamic virtual reality dataset for aerial threat object detection

“…8 For effective training of machine learning models used in UAV detection, the environment in which the UAVs operate should provide variation in landscape, weather conditions, time of day, and terrain features. 9 Additionally, machine learning models require specific dataset formats for training. 10 Output data should follow conventional ML formats, including label and bounding box information of all aerial objects rendered.…”