Online Learning in a Chemical Perceptron

Journal of the Royal Society, Interface / the Royal Society, 2014

Learning and adaptive behaviour are fundamental biological processes. A key goal in the field of bioengineering is to develop biochemical circuit architectures with the ability to adapt to dynamic chemical environments. Here, we present a novel design for a biomolecular circuit capable of supervised learning of linear functions, using a model based on chemical reactions catalysed by DNAzymes. To achieve this, we propose a novel mechanism of maintaining and modifying internal state in biochemical systems, thereby advancing the state of the art in biomolecular circuit architecture. We use simulations to demonstrate that the circuit is capable of learning behaviour and assess its asymptotic learning performance, scalability and robustness to noise. Such circuits show great potential for building autonomous in vivo nanomedical devices. While such a biochemical system can tell us a great deal about the fundamentals of learning in living systems and may have broad applications in biomedic...

2000

Abstract—Today’s computers are built up from a minimal set of standard pattern recognition opera- tions. Logic gates, such as NAND, are common ex- amples. Biomolecular materials offer an alternative approach, both in terms of variety and context sensi- tivity. Enzymes, the basic switching elements in bio- logical cells, are notable for their ability to discrimi- nate specific molecules in a

Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 2015

One contribution of 13 to a Theo Murphy meeting issue 'Heterotic computing: exploiting hybrid computational devices' .

The behaviour of pulses of Belousov-Zhabotinski (BZ) reaction-diffusion waves can be controlled automatically through machine learning. By extension, a form of chemical network computing, i.e., a massively parallel non-linear computer, can be realised by such an approach. In this initial study a lightsensitive sub-excitable BZ reaction in which a checkerboard image comprising of varying light intensity cells is projected onto the surface of a thin silica gel impregnated with tris(bipyridyl) ruthenium (II) catalyst and indicator is used to make the network. As a catalyst BZ solution is swept past the gel, pulses of wave fragments are injected into the checkerboard grid resulting in rich spatio-temporal behaviour. This behaviour is shown experimentally to be repeatable under the same light projections. A machine learning approach, a learning classifier system, is then shown able to direct the fragments to an arbitrary position through dynamic control of the light intensity within each cell in both simulated and real chemical systems.

2009

Biological olfaction outperforms chemical instrumentation in specificity, response time, detection limit, coding capacity, time stability, robustness, size, power consumption, and portability. This biological function provides outstanding performance due, to a large extent, to the unique architecture of the olfactory pathway, which combines a high degree of redundancy, an efficient combinatorial coding along with unmatched chemical information processing mechanisms. The last decade has witnessed important advances in the understanding of the computational primitives underlying the functioning of the olfactory system. In this work, the state of the art concerning biologically inspired computation for chemical sensing will be reviewed. Instead of reviewing the whole body of computational neuroscience of olfaction, we restrict this review to the application of models to the processing of real chemical sensor data.

Synthetic biological circuits are promising tools for developing sophisticated systems for medical, industrial, and environmental applications. So far, circuit implementations commonly rely on gene expression regulation for information processing using digital logic. Here, we present a new approach for biological computation through metabolic circuits designed by computer-aided tools, implemented in both whole-cell and cell-free systems. We first combine metabolic transducers to build an analog adder, a device that sums up the concentrations of multiple input metabolites. Next, we build a weighted adder where the contributions of the different metabolites to the sum can be adjusted. Using a computational model trained on experimental data, we finally implement two four-input perceptrons for desired binary classification of metabolite combinations by applying model-predicted weights to the metabolic perceptron. The perceptron-mediated neural computing introduced here lays the groundw...

Abstract. All information processing systems found in living organisms are based on chemical processes. Harnessing the power of chemistry for computing might lead to a new unifying paradigm coping with the rapidly increasing complexity and autonomy of computational systems. Chemical computing refers to computing with real molecules as well as to programming electronic devices using principles taken from chemistry.

Lecture Notes in Computer Science, 2012

Learning and generalisation are fundamental behavioural traits of intelligent life. We present a synthetic biochemical circuit which can exhibit nontrivial learning and generalisation behaviours, which is a first step towards demonstrating that these behaviours may be realised at the molecular level. The aim of our system is to learn positive real-valued weights for a real-valued linear function of positive inputs. Mathematically, this can be viewed as solving a non-negative least-squares regression problem. Our design is based on deoxyribozymes, which are catalytic DNA strands. We present simulation results which demonstrate that the system can converge towards a desired set of weights after a number of training instances are provided.

Angewandte Chemie International Edition in English, 1993

The capabilities of the human brain have always fascinated scientists and led them to investigate its inner workings. Over the past 50 years a number of models have been developed which have attempted to replicate the brain's various functions. At the same time the development of computers was taking a totally different direction. As a result, today's computer architectures, operating systems, and programming have very little in common with information processing as performed by the brain. Currently we are experiencing a reevaluation of the brain's abilities, and models of information processing in the brain have been translated into algorithms and made widely available. The basic building‐block of these brain models (neural networks) is an information processing unit that is a model of a neuron. An artificial neuron of this kind performs only rather simple mathematical operations; its effectiveness is derived solely from the way in which large numbers of neurons may be ...

International journal of molecular sciences, 2018

DNA molecular machines have great potential for use in computing systems. Since Adleman originally introduced the concept of DNA computing through his use of DNA strands to solve a Hamiltonian path problem, a range of DNA-based computing elements have been developed, including logic gates, neural networks, finite state machines (FSMs) and non-deterministic universal Turing machines. DNA molecular machines can be controlled using electrical signals and the state of DNA nanodevices can be measured using electrochemical means. However, to the best of our knowledge there has as yet been no demonstration of a fully integrated biomolecular computing system that has multiple levels of information processing capacity, can accept electronic inputs and is capable of independent operation. Here we address the question of how such a system could work. We present simulation results showing that such an integrated hybrid system could convert electrical impulses into biomolecular signals, perform ...