The ideas introduced in this book explore the relationships among big data,
machine learning and granular computing. In many studies, machine learning has
been considered as a powerful tool of big data processing. The relationship between
big data and machine learning is very similar to the relationship between resources
and human learning. In this context, people can learn from resources to deal with
new matters. Similarly, machines can learn from big data to resolve new problems.
However, due to the vast and rapid increase in the size of data, learning tasks have
become increasingly more complex. In this context, traditional machine learning
has been too shallow to deal with big data sufficiently, so granular computing
concepts are used in this book to advance machine learning towards the shift from
shallow learning to deep learning (in its broader sense).
The focus of this book is on the development and evaluation of granular computing
based machine learning approaches in terms of classification accuracy. In this context,
the authors consider traditional machine learning to be of single-granularity and the
proposal of granular computing based machine learning is aimed at turning
single-granularity learning into multi-granularity learning. In particular, the authors
proposed the following transformations: (a) supervised learning to semi-supervised
learning, (b) heuristic learning to semi-heuristic learning, (c) single-task learning to
multi-task learning, (d) discriminative learning to generative learning and (e) random
data partitioning to semi-random data partitioning. In addition, the authors also explore
how to achieve in-depth evaluation of attribute-value pairs towards induction of
high-quality rules, in the setting of multi-granularity learning.
