INTELLIGENT SYSTEMS
- Overview
- Assessment methods
- Learning objectives
- Contents
- Bibliography
- Delivery method
- Teaching methods
- Contacts/Info
The course assumes that students have a background acquired on a Bachelor's Degree in Computer Science. Specifically, students will be expected to be familiar with Vector and Matrix Calculation and Statistics
The objective of the evaluation procedure is the assessment of the acquired knowledge and ability listed above.
The evaluation procedure is based on two methods as follows:
1)written examination lasting 2 hours. The test consists of 5 questions, some of these entirely of theory, some including numerical exercises similar to those solved during the lessons and detailed in the course slides (6 points available for each question).
2) Report regarding an in-depth exploration of a topic covered in the course with additional material chosen with the supervision of the teacher. The activity can be performed in small groups limited to a maximum of three members.
The separate assessment elements contribute towards the final mark according to the following rule:
• Written text 70%
• Project 30%
Final mark is expressed in thirtieths.
After correction, the student is invited for a review of the written test and the report; if case the final mark is augmented.
The course provides a broad coverage of intelligent systems solving pattern recognition problems. Theoretical concepts in intelligent systems and techniques relevant in real-life applications will be illustrated.
The student will be able to:
1. Know the main objectives and areas of the Artificial Intelligence, of Machine Learning and Pattern Recognition, with the ability to identify the potentialities of the intelligent techniques and the relationships with other disciplines
2. Know the basic concepts of the automated classifiers based on machine learning approaches and the conditions for their applicability
3. Know most relevant feature extraction and selection techniques
4. Know statistical techniques and their limits and strengths, with the ability to appropriately choose the proper technique in specific contexts
5. Know basic principle of neural computing and the distinguishing characteristics
6. Know supervised feed forward neural models, with the ability to apply heuristics in the choice of network topology, to check the condition of overfitting, to evaluate performances in training and test.
7. Know Flat and Hierarchical Clustering with the ability to configure and apply these methods in specific contexts
8. Know Competitive and Self-Organizing approaches
9. Know basic concepts of Fuzzy Logic with the ability to design fuzzy rule based systems
10. Know accuracy evaluation metrics
11. Know basic concepts of the following application domains: Image Classification, Text Categorization, Biomedical Data Analysis
It is also expected that students develop communicative skills through open discussion and autonomous assessment in the choice of the proper technique to solve problems of recognition and /or automatic classification of multidimensional data in several domains.
Students will acquire also knowledge of the relevant Machine learning and Pattern Recognition terminology.
The acquisition of knowledge and expected skill is developed along the entire course that includes the topics listed below.
1)Introduction to Artificial Intelligence and Pattern Recognition: Historical Perspective, State of the Art of methods and applications (4h- Course Objective 1)
2) Design of a supervised classifier; Basic principles of learning by example; basic concepts of multidimensional pattern analysis (4 h- Course Objective 2)
3)Feature extraction and selection:
• Principal Component Analysis
• Information Gain
• Statistical evaluation of features
• Sequential Forward Selection,
• Design of Multidimensional Patterns in Text Classification, Image Recognition and in Content-Based Retrieval
(8h Course Objective 3)
4) Automated Classification: conventional basic approaches
• Minimum distance classifier
• Deterministic distances
• Bayes’s rule
• Maximum likelihood classifier
• K-Nearest Neighborhood classifier
• Parallelepiped Method
(10h- Course Objective 4)
5)Neural Networks
• Introduction, taxonomy
• Basic principle of neural computing
(3h- Course Objective 5)
6)Feed forward Neural Models:
• Perceptron,
• Fixed increment rule, Delta rule,
• Limits of the perceptron model: XOR problem
• Multilayer Perceptron: topology
• back-propagation learning rule and derived algorithms,
• Example of Applications
(10 h- Course Objective 6)
7) Clustering
• Introduction to Clustering
• K-means Clustering algorithm
• Agglomerative Hierarchical Clustering: Single linkage, Complete linkage
(5h- Course Objective 7)
8) Competitive Neural Learning
• Self Organizing networks;
• Kohonen Networks
(7h- Course Objective 8)
9) Soft Computing
• Fuzzy Sets Theory
• Approximate Reasoning
• Fuzzy C-means algorithm
(6h- Course Objective 9
.
10) Introduction to Deep Learning techniques
(5h- Course Objective 6)
11) Evaluation metrics
• Confusion Matrix
• Precision, Recall, Jaccard and Dice evaluation indexes
• Examples of application of evaluation metrics in Image Classification, Medical Informatics and Information Retrieval domains
(5h- Course Objective 10)
12) Design of Intelligent Systems, Examples in application domains
(5h- Course Objectives 11)
Material available on the e-learning website:
Slides of the lessons, Published scientific papers, links to websites with demo and tutorials
Examples of written texts with solutions and examples of projects
Textbook:
- R.O. Duda, P.E. Hart, and D.G. Stork, Pattern Classification, New York: John Wiley & Sons, 2001
- T. Mitchell, Machine Learning, McGraw Hill, 1997.
- C.M. Bishop, Neural Networks for Pattern Recognition, Oxford University Press, Inc. New York, NY, USA, 1995
Lessons (72 hours)
Lessons deal with the overall set of topics listed above using conceptual, formal descriptions and with the support of Demo and on line resources. Open discussions with students during lessons are encouraged.
Office hours:
Office hours is agreed by e-mail: name.surname@uninsubria.it; the teacher responds only to e-mail sent from student.uninsubria.it
Professors
Borrowers
-
Degree course in: PHYSICS
-
Degree course in: MATHEMATICS
-
Degree course in: MATHEMATICS