COURSE DESCRIPTION
Theory of neural network models. Relation to biological models. Examples of known models and possible applications. Digital vs. analog approaches and building blocks.
PRE-REQUISITE : EE 3353 and EE 3384 , each with grade of "C" or better.
POST-REQUISITE :
COURSE OUTCOMES : Students completing EE 4365 will be able to:
- Identify different neural network architectures, their limitations and appropriate learning rules for each of the architectures. (C)
- Select appropriate neural network architectures for a given application. (i.e. they shall recognize the class of applications and relate it to specific architectures). (C)
- Verify the classic linear separability problem that exists for single layer networks, demonstrate and explain how adding a hidden layer solves the problem. (I)
- Design and implement a neural network simulation (with two modes of operation: learning and processing) using a high-level language. (I)
- Apply Boolean algebra to verify the analogy between a neural network and a digital logic circuit. Additionally, signal and systems concepts will be used to verify the analogy between a neuron and feedback controller, as well as the analogy between a neuron and an infinite impulse response filter. (D)
- Make a presentation either detailing a current application of neural nets or summarizing a current article in the field. (D)
PREVIOUS COURSE NUMBER: EE3465
RESPONSIBLE COMMITTEE: Computer Engineering
PREVIOUSLY OFFERED
| FALL | SPRING | SUMMER |
| all | - | - |
TO BE OFFERED (Tentative)
| FALL | SPRING | SUMMER |
| 2001 | - | - |
| 2002 | - | - |
| 2003 | - | - |