Training a Carbon-Nanotube/Liquid Crystal Data Classifier Using Evolutionary Algorithms
Evolution-in-Materio uses evolutionary algorithms (EA) to exploit the physical properties of unconfigured, physically rich materials, in effect transforming them into information processors. The potential of this technique for machine learning problems is
- PDF / 464,127 Bytes
- 12 Pages / 439.37 x 666.142 pts Page_size
- 87 Downloads / 156 Views
Abstract. Evolution-in-Materio uses evolutionary algorithms (EA) to exploit the physical properties of unconfigured, physically rich materials, in effect transforming them into information processors. The potential of this technique for machine learning problems is explored here. Results are obtained from a mixture of single walled carbon nanotubes and liquid crystals (SWCNT/LC). The complex nature of the voltage/current relationship of this material presents a potential for adaptation. Here, it is used as a computational medium evolved by two derivative-free, population-based stochastic search algorithms, particle swarm optimisation (PSO) and differential evolution (DE). The computational problem considered is data classification. A custom made electronic motherboard for interacting with the material has been developed, which allows the application of control signals on the material body. Starting with a simple binary classification problem of separable data, the material is trained with an error minimisation objective for both algorithms. Subsequently, the solution, defined as the combination of the material itself and optimal inputs, is verified and results are reported. The evolution process based on EAs has the capacity to evolve the material to a state where data classification can be performed. PSO outperforms DE in terms of results’ reproducibility due to the smoother, as opposed to more noisy, inputs applied on the material.
1
Introduction
Unconventional computing aims at investigating methods for designing systems able to perform a computation in different ways than the current paradigm. One such direction of research is evolution in materio (EIM) [10], which is concerned with computing performed directly by the materials. EIM focuses on the underlying properties of the materials aiming at exploring and exploiting them in such a way so that they are brought to a computation inducing state. Contrary to traditional computing with metal-oxide-silicon-field-effect-transistor (MOSFET) technology, where everything is designed, produced and programmed very carefully, EIM uses a bottom up approach where computation is performed by the material without having explicit knowledge of its internal properties. c Springer International Publishing Switzerland 2016 M. Amos and A. Condon (Eds.): UCNC 2016, LNCS 9726, pp. 130–141, 2016. DOI: 10.1007/978-3-319-41312-9 11
Training a Carbon-Nanotube/Liquid Crystal Data
131
The idea of EIM can be found in early work of Pask [17] concerned with growing an electrochemical ear. In more recent work [22], observations were made when evolutionary algorithms were used for designing electrical circuits on field-programmable-gate-arrays (FPGAs). The resulting circuit topologies were influenced by the material of the board used. Because of feedback provided by the iterative nature of stochastic optimisation interacting with the material, identified solutions were based on the specific FPGA’s properties that were unaccounted for during the board’s design. EIM replaced the FPGAs with material sys
Data Loading...
