As an introduction we will define some multidisciplinary terminology, consider our motivations, and cover some relevant academic activity as well as research publications.
1.1 Relevant terminology and related bio-inspired technologies
Biomimetic implies the mimicry of biology; note the word mime embedded in the term. In recent years biomimicry seems to be used more and more for sensory systems applications (our interest), while biomimetics implies molecular-level mimicry. This text is focused on biologically-inspired paradigms used for sensory systems and the signal processing that goes with such systems. The subject is sensory systems and not the research associated with mimicking organic chemistry, muscle tissue, etc. In this text cursory descriptions of biological phenomena will be followed by electronic sensor designs and the signal processing algorithms that emulating such phenomena for useful technological application. Bioprincipic is a similar term used recently implying the mimicry of biological principles.
Biometric implies measuring biological features unique to an individual to determine the identity of the person. For example, authentication can be granted based on a pattern matching of a fingerprint, scanned iris image, or recorded voice pattern. This could be used for building and computer security purposes.
Biomedical means the branch of medicine associated with survival in stressing environments. Bionic means enhancing normal biological capability with electronic or mechanical devices [Webster].
Bioinformatics is used to describe computer applications of extracting information about biological phenomena, primarily in the field of molecular biology. Bioinformatics is more formally defined as "The collection, classification, storage, and analysis of biochemical and biological information using computers especially as applied to molecular genetics and genomics." [Webster].
Anatomy and Physiology imply structure and function, respectively. Scientists from many disciplines often organize their thoughts in similar ways. However, until there is a reason to communicate across disciplines, the terminology in each tends to develop into differently. Table 1 is an observation of the separation of phenomena into physical and abstract categories.
Table 1. Different Terminology, Similar Concepts
Genetic Algorithms refers to computational methods inspired by genetics. A genetic algorithm may consist of improving on an existing solution (or one chosen initially at random) based on an evaluation of fitness representing the problem solution. Improved solutions may be derived from fitness evaluation and genetic operators representing mutation and crossover.
Evolutionary Computation refers to the computational methods mimicking natural evolutionary forces.
Neural Networks is used to refer to networks of computational elements that process information in an analogous way to biological neuronal networks. Both natural and artificial neural networks perform a nonlinear transform on an aggregation of many weighted input signals. There are many artificial “neural network” paradigms (ANN's) that include many ideas not found in biological neuronal networks, although the general concept has its original inspiration from biology.
Nevertheless, most ANN variations have these features in common with natural neural networks:
- A summation of many inputs, each weighted differently based on learned examples
- A non-linear output mapping function follows the summation
- massively parallel
- distributive processing
The application of ANN’s to various engineering applications has grown into an academic field of its own, with separate texts and courses dedicated to the field. Further study of neural networks is reserved for courses and texts dedicated to this subject.