Page 11 - Data Science Algorithms in a Week
P. 11
viii Luis Rabelo, Sayli Bhide and Edgar Gutierrez
first and then the basics of ensemble learning are given. Finally, the chapter concludes
with a summary of the novel progresses in unsupervised learning.
• Deep Learning and a Complex Application in Parallel Distributed Simulation:
is introduced in the chapter by Edwin Cortes and Luis Rabelo entitled “Using Deep
Learning to Configure Parallel Distributed Discrete-Event Simulators.” The authors
implemented a pattern recognition scheme to identify the best time management and
synchronization scheme to execute a particular parallel discrete simulation (DES)
problem. This innovative pattern recognition method measures the software complexity.
It characterizes the features of the network and hardware configurations to quantify and
capture the structure of the Parallel Distributed DES problem. It is an innovative research
in deep belief network models.
• Autonomous Systems: The area of autonomous systems as represented by
autonomous vehicles and deep learning in particular Convolutional Neural Networks
(CNNs) are presented in the chapter “Machine Learning Applied to Autonomous
Vehicles” by Olmer García and Cesar Díaz. This chapter presents an application of deep
learning for the architecture of autonomous vehicles which are a good example of a
multiclass classification problem. The authors argue that the use of AI in this domain
requires two hardware/software systems: one for training in the cloud and the other one in
the autonomous vehicle. This chapter demonstrates that deep learning can create
sophisticated models which are able to generalize with relative small datasets.
• Genetic Algorithms & Support Vector Machines: The utilization of Genetic
Algorithms (GAs) to select which learning parameters of AI paradigms can actually assist
researchers in automating the learning process is discussed in the chapter “Evolutionary
Optimization of Support Vector Machines Using Genetic Algorithms”. Fred Gruber uses
a GA to find an optimized parameter set for support vector machines. GAs and cross
validation increase the generalization performance of support vector machines (SVMs).
When doing this, it should be noted that the processing time increases. However, this
drawback can be reduced by finding configurations for SVMs that are more efficient.
• Texture Descriptors for the Generic Pattern Classification Problem: In the
chapter “Texture Descriptors for the Generic Pattern Classification Problem”, Loris
Nanni, Sheryl Brahnam, and Alessandra Lumini propose a framework that employs a
matrix representation for extracting features from patterns that can be effectively applied
to very different classification problems. Under texture analysis, the chapter goes through
experimental analysis showing the advantages of their approach. They also report the
results of experiments that examine the performance outcomes from extracting different
texture descriptors from matrices that were generated by reshaping the original feature
vector. Their new methods outperformed SVMs.
• Simulation Optimization: The purpose of simulation optimization in predicting
supply chain performance is addressed by Alfonso Sarmiento and Edgar Gutierrez in the
chapter “Simulation Optimization Using a Hybrid Scheme with Particle Swarm