We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.ET

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Emerging Technologies

Title: Training DNA Perceptrons via Fractional Coding

Abstract: This paper describes a novel approach to synthesize molecular reactions to train a perceptron, i.e., a single-layered neural network, with sigmoidal activation function. The approach is based on fractional coding where a variable is represented by two molecules. The synergy between fractional coding in molecular computing and stochastic logic implementations in electronic computing is key to translating known stochastic logic circuits to molecular computing. In prior work, a DNA perceptron with bipolar inputs and unipolar output was proposed for inference. The focus of this paper is on synthesis of molecular reactions for training of the DNA perceptron. A new molecular scaler that performs multiplication by a factor greater than 1 is proposed based on fractional coding. The training of the perceptron proposed in this paper is based on a modified backpropagation equation as the exact equation cannot be easily mapped to molecular reactions using fractional coding.
Comments: Proc. 2019 Asilomar Conference on Signals, Systems and Computers
Subjects: Emerging Technologies (cs.ET); Quantitative Methods (q-bio.QM)
DOI: 10.1109/IEEECONF44664.2019.9048931
Cite as: arXiv:1911.07110 [cs.ET]
  (or arXiv:1911.07110v2 [cs.ET] for this version)

Submission history

From: Keshab Parhi [view email]
[v1] Sat, 16 Nov 2019 22:48:28 GMT (1148kb,D)
[v2] Wed, 8 Jan 2020 23:06:02 GMT (1147kb,D)

Link back to: arXiv, form interface, contact.