Search    Print

Digital Adaptive Circuits and Learning Systems
Digital Adaptive Circuits and Learning Systems
Stefano Squartini

Seat Ingegneria
A.A. 2016/2017
Credits 9
Hours 72
Period I
Language ENG

Prerequisites
Linear Algebra, Electrical Circuit Theory, Circuits and Algorithms ofr Digital Signal Processing

Learning outcomes
KNOWLEDGE AND UNDERSTANDING:
The student is expected to know and understand advanced Digital Signal Processing (DSP) techniques, with special focus to the analysis, synthesis and implementation of adaptive discrete-time circuits and algorithms, both linear and nonlinear, including the artificial neural networks.
CAPACITY TO APPLY KNOWLEDGE AND UNDERSTANDING:
The student is expected to acquire the ability of applying the advanced DSP techniques discussed during the lectures in specific audio processing problems. On purpose, the student will be asked to accomplish a project targeted to the implementation of an algorithm by means of suitable SW tools (both on PC and embedded platforms)
TRANSVERSAL SKILLS:
The student is expected to design advanced DSP algorithm and to accomplish suitable analytical studies by exploiting theoretical models, computer simulations and laboratory experiments. Moreover, the student is expected to be able to critically evaluate the data obtained from experiments and simulations, to draw conclusions and take decisions with the objective to optimize the performance of proposed solutions.

Program
Review of basic DSP concepts. Review of Estimation Theory concepts Optimal FIR filters. Linear prediction and modern spectral analysis. Adaptive FIR filter in both time and frequency domains. IIR adaptive filters. Adaptive filter relevant applications. Static and Dynamic Neural Networks. Neural Networks applications Advanced Algorithms for Computational Audio Processing. Implementation of adaptive algorithms and neural networks in MATLAB Real-time implementation of adaptive algorithms on Digital Signal Processors

Development of the examination
LEARNING EVALUATION METHODS
The learning evaluation methodology consists in the presentation and discussion of a technical report related to a project focused on the Advanced Digital Signal Processing and Computational Intelligence presented during the lectures, with special focus on Audio Processing applications, and to be develoepd by means of suitable SW tools (running on PC and/or Embedded Platforms). The project is proposed by the teacher in agreement with the student's preferences and it can be fulfilled also in groups of maximum two people. The student can also propose something on the basis of his/her interests: the teacher will carefully evaluate the suitability of student's suggestion in relationship with the course contents, and will also calibrate the implementation aspects before finalizing the project proposal. The student has 6 months to fulfill the work and present it, starting from the date in which the project proposal is given to him/her.

LEARNING EVALUATION CRITERIA
The student is required to show an adequate comprehension of the concepts discussed during the lectures and to be able to apply them in an autonomous way to the fulfillment of the assigned project. The student is also asked to explain in a rigorous and synthetic way the technical report related to the project.

LEARNING MEASUREMENT CRITERIA
During the project fulfillment and related final discussion, the learnign measurement criteria will be focused on the ability to apply the advanced Digital Signal Processing and Computational Intelligence methodologies and techniques, studied during the lectures, and in relationship with the project objectives. Moreover, the capability to autonomously face the issues raising during the project development will be evaluated. These issues will be related to the analysis of the technical requirements, the implementation on the selected HW/SW platform and the critical assessment of performance. Finally, the capability to carry out suitable analytical studies by exploiting theoretical models, computer simulations and laboratory experiments, will be also evaluated.

FINAL MARK ALLOCATION CRITERIA
The test is evaluated with a score within the 18-30 range. To pass the exam, the student is required to show his/her ability in analyzing the problems related to the proposed project, by using the technical background acquired during the lectures, and develop suitable algorithms for their solutions. The maximum score is reached when the student show to be able to work in an autonomous and costructive way to face the technical issues experienced during the project fulfillment, and to motivate, by means of adequate experimental tests, the functional properties of the algorithms. Honours are given to students who show to have a relevant scientific rigour in addressing the faced issues and a certain brightness in exposing their answers to specific theoretical questions and in discussing the final project.

Recommended reading
-P. M. Clarkson, Optimal and Adaptive Signal Processing, CRC Press, 2000; 2 -S.Haykin, Neural Networks, IEEE Press, 1994 (o edizioni successive); 3- D. Reay, Digital Signal Processing and Applications with the OMAP - L138 eXperimenter, Wiley and Sons, 2012; 4- Teacher’s material available at the website https://lms.univpm.it/, and specifically at the pages related to this course.

Courses

• Ingegneria Elettronica (Corso di Laurea Magistrale (DM 270/04))