First, some background; Neural nets are by design black box. It is less important to understand the problem you are solving when designing a neural network than it is when writing a deterministic algorithm to solve it directly. With that in mind, you probably don't need to learn about "intrusion detection systems using neural networks", but would probably benefit more from learning about neural networks and intrusion systems separately.
I will leave it to you to find texts on intrusion detection systems, but would recommend reading the following to get started on what neural networks are, and how they work:
Neural Networks - A Systematic Introduction
If you think you have understood the basis of neural networks conceptually, you will want to learn a programming language. Your options diverge somewhat at this point, but I would suggest that if you want to learn neural nets from an academic perspective and want to have more control over the design and guts of the program, you would probably benefit most from learning C++. There is a wealth of knowledge on the topic of learning C++ online. In fact, probably the most popular page on this website is dedicated to that topic:
The Definitive C++ Book Guide and List
Once you understand neural network fundamentals and C++, the world is your oyster! If you're feeling adventurous, have a look at Kenneth Stanley's NEAT algorithm. The source code will teach you a lot about neural net algorithms.
From here to creating a learning machine that understands intrusion attempts is almost trivial from a programming perspective. You really just need to get the data, which may be really easy or really hard, but your supervisor should be able to help you find data sources on which to train the network once you reach this point.
Good luck!