|Teoria||5||I semestre||Marco Cristani|
|Laboratorio||1||I semestre||Marco Cristani|
The course aims to provide: i) advanced techniques of statistical recognition and machine learning, as discriminative and neural classifiers (deep learning); ii) advanced techniques for the programming of professional code for classification in industrial environments; iii) knowledge of classification problems of the industrial world, and techniques usually used for their resolution.
At the end of the course the student must demonstrate to be able to: i) understand if a classification problem can be solved without the existing technologies; ii) understand what type of learning algorithm should be used for training a classifier.
Furthermore, he / she must demonstrate that he / she has the ability to apply the acquired knowledge: i) identifying what type of classifier or recognizer should be used in response to a given problem; iii) understanding that the machine learning strategy must be implemented according to the number of training data available; iii) understanding the complexity of the problem of recognition in computational terms; iv) being able to write professional software that recognizes real data, possibly modifying it in relation to the problem under examination.
This knowledge will allow the student to understand that measures of error and performance must be taken into account given a specific problem under consideration. Furthermore, this knowledge will enable the student to continue his or her studies autonomously in the context of automatic learning or recognition.
The course presents a series of state-of-the-art topics in the field of recognition. Each topic will be explained through updated articles together with the lesson slides. The following books are suggested as a reference:
- Christopher M. Bishop. 2006. Pattern Recognition and Machine Learning (Information Science and Statistics). Springer-Verlag New York, Inc., Secaucus, NJ, USA.
- Ian Goodfellow, Yoshua Bengio, and Aaron Courville. Deep learning. MIT Press, 2016.
- Classification validation tools: Confusion matrix and derivative measurements, ROC and CMC curves, average precision, average quadratic error, label correlation, grading and regression measures
- Kernel machines, Support Vector Machines
- VLFeat for object recognition: Dense object recognition through multiclass discriminatory models
- Dense classification features as bag of words
- Shape descriptors for object tracking: B-spline and Condensation
- Deep learning in Tensorflow: Multinomial Logistic Classifier, Neural Networks, Convolutional Neural Network
The exam involves the discussion of a code project, which proposes a solution to an industrial classification problem. The final score will depend on the classification figure of merits achieved by the classifier and the theoretical motivations that prompted the student to choose a particular algorithm.