Experiments by DeepMind playing Go to beat humans have put neural networks into common news threads. These algorithms are being used for image and speech analysis, and they have shown a lot of promise in predicting and automating some systems in engineering and computing.

Neural networks have several variations. For example, the feed-forward neural network is good for clustering and image analysis, whereas the recurrent neural network is more suited to speech or temporal data analysis. This book discusses temporal neural networks and how they impact artificial intelligence (AI).

This book could be part of a course studying advanced topics in neural networks. It gives an in-depth description of how biological neurons work and how they inspire algorithms. The taxonomy of the variant models used by researchers working in the field is useful in showing how work changes from one model to the next. This presents a good review of existing techniques and what differences exist between them.

But what are temporal neural networks? These are different from other neural networks (which are triggered by activation functions); these are self-timed and use spikes to trigger responses. The responses are then analyzed as time-based results, allowing the activation function to produce more variable results than the traditional neural networks. These algorithms also allow different time scales, such as local versus global, to influence the neuron spikes in the algorithms.

The book goes into the complexity of coding these algorithms. Discussing it from a control theory perspective, the authors make it easy to understand, especially through the use of logic gates and simple rules. Further, an application of the MNIST database shows how well image recognition can be done using temporal neural networks.

The book chapters are well structured, dedicated to specific topics in temporal neural networks. They carry examples of how these can be applied to various problems. However, when searching I could not find a website or any accompanying resources to test out temporal neural networks. The technique does have potential but would need more examples to test out in code and experiments.

Overall, the book will be very useful for students, especially beginners, to analyze how temporal neural networks could benefit their work. The basics in this book can lead to more complicated applications of the algorithms.