Introduction to Deep learning and Applications to Undergraduate Students
Machine learning allows computers to learn what comes naturally to humans and animals. The method learns from experience. It employs the use of computational methods to learn an insight directly from data without using a predetermined equation as a model. This method allows the algorithms to adaptively improves their performance as the number of input data available for learning increases.The machine learning algorithms look for natural patterns in data that produces insight and provide better decisions making and future predictions. They are used in different applications. For example, medical diagnostics, stock trading, weather forecasting, robotics, automobiles, etc. As the volume of data increases, machine learning has become particularly important for providing solutions in the following areas:
1. Computational finance: in predicting the stock prices, and online trading of stocks.
2. Image processing and computer vision: use in object detection, tracking and recognition applications.
3. Computational biology: guse in applications such as tumor detection, drug discovery and DNA sequencing.
4. Energy production: use for smart metering and future energy processing.
5. Automotive, aerospace, and manufacturing: use in providing predictive maintenance.
In this classroom activity, the students will learn how to develop machine learning and deep learning models and how to choose between supervised and unsupervised techniques to solve problems using MATLAB.
 MathWorks, Introducing Machine learning with MATLAB, Ebook, MatWorks Inc., 80789V00, pp. 1-63.
 Judith Hurwitz and Daniel Kirsch, IBM Machine Learning for Dummies, John Wiley & Sons, Inc. 111 River St. Hoboken, NJ 07030-5774, 2018, pp. 8-12.
 MathWorks, Introducing Deep learning with MATLAB, Ebook, MatWorks Inc., 80789V00, pp. 1-13.
 Machine Learning, Tutorial Point, Easy learning, available at: http//www.tutorialsponit.com.
 Tom M. Mitchell, Machine Learning, McGraw-Hill Inc. March 1, 1997, ISBN: 0070428077, pp. 13-30.
The main goal of this classroom activity is to provide students with hands-on experience on machine and deep learning models and applications using MATLAB programming and application skills covered in the classroom. The activity will expose the students with a real-life demonstration of machine and deep learning algorithms and applications in different field. The students will learn the following:
1. What is machine and deep learning?
2. How Machine Learning Works?
3. How Do You Select the Best Algorithm to Use?
4. When Should You Use Machine Learning?
5. What are the Machine Learning Challenges?
6. The Machine Learning Workflow
7. What Makes Deep Learning State-of-the-Art?
8. How A Deep Neural Network Learns?
9. An Example Using AlexNet Dataset?
10. How to Retraining an Existing Network?
The prerequisite MATLAB skills needed for this classroom activity are as follows:
1. Constant / variables / expressions
2. Functions / Subroutines
3. MATLAB plotting, coloring and visualization
4. Arrays and vector indexing
5. Objects / Instance
Context for Use
I used the teaching activity as a classroom activity in a 400-level undergraduate engineering students' course titled, "EA413 computer programming for engineers (MATLAB programming)". The students have covered the basics of engineering problem solving techniques, vector analysis, computational models as applied in mathematics, basic introduction to MATLAB features and built-in functions as the prerequisite to this classroom activity. However, the class size is too large, about three hundred and fifty (350) students are taking the course. Students from the following engineering discipline are taking the course, these include: electrical / electronics, mechanical / production, automobile, civil, mechatronics, agric, petroleum, chemical and physics engineering. The classroom activity includes a practical demonstration using MATLAB codes and a real-life application of machine and deep learning algorithms. The duration of this class activity including the practical demonstration is two hours (2 Hours). This makes the teaching and learning easier for the students and develop interest in machine and deep learning applications using MATLAB computational skills.
Description and Teaching Materials
The main emphasis of this classroom activity is to provide students with hands-on experience on machine and deep learning models and algorithms using MATLAB programming and problem-solving skills. The activity will expose the students with a real-life demonstration of machine and deep learning techniques and applications. The activity is to be conducted as in-class activity. The activity takes about two (2) hours. As mentioned earlier in the summary section, the number of students taking the course is large. Therefore, the first one hour will be devoted to teaching the basic principles and concept as provided in the activity teaching lecture note. In the second hour, the student will be divided into groups and provided with the description of the problem statement.The last twenty minutes of the second hour will be used to check the correctness of the implementation, result accuracy and difficulties encountered by the students.
Activity lecture note: lntroduction to Machine and Deep Learning
Problem source codes: MATLAB file (Source Codes.pdf)
Program Source Code (link down)
Teaching Notes and Tips
The students should cover the basic introduction to MATLAB and built-in functions, arrays and vector indexing, loops, plotting and visualization as a prerequisite to this activity. The instructor should demonstrates the basic introduction to machine and deep learning and applications in real life situations and use MATLAB as a tool for the design and implementation. Sometimes the students may encounter a difficulty in verifying the accuracy of their results. To test the accuracy of the MATLAB program, the student should modify and run the program using different parameter values and notice the change in the model accuracy.
In order to ensure that the students have met the goals of this classroom activity, the following questions should be answered?
1. What type of data are you planning to work with?
2. What insights do you want obtained from the data?
3. How to applied the obtained insights in different applications?
If the above questions are well answered, we can rest assured that the classroom activity has worked quite well. Then, the student should be introduced to the possible application of the concepts in different application domains, such as aerospace, big data, automobiles, medical, etc.