Fully parallel write/read in resistive synaptic array for accelerating on-chip learning.

A neuro-inspired computing paradigm beyond the von Neumann architecture is emerging and it generally takes advantage of massive parallelism and is aimed at complex tasks that involve intelligence and learning. The cross-point array architecture with synaptic devices has been proposed for on-chip implementation of the weighted sum and weight update in the learning algorithms. In this work, forming-free, silicon-process-compatible Ta/TaOx/TiO2/Ti synaptic devices are fabricated, in which >200 levels of conductance states could be continuously tuned by identical programming pulses. In order to demonstrate the advantages of parallelism of the cross-point array architecture, a novel fully parallel write scheme is designed and experimentally demonstrated in a small-scale crossbar array to accelerate the weight update in the training process, at a speed that is independent of the array size. Compared to the conventional row-by-row write scheme, it achieves >30× speed-up and >30× improvement in energy efficiency as projected in a large-scale array. If realistic synaptic device characteristics such as device variations are taken into an array-level simulation, the proposed array architecture is able to achieve ∼95% recognition accuracy of MNIST handwritten digits, which is close to the accuracy achieved by software using the ideal sparse coding algorithm.

Modeling of gene regulatory network dynamics using threshold logic.

Gene regulation modeling is one of the most active research topics in systems biology. The aim of modeling gene regulation is to understand how individual genes function and interact with each other to create complex biological phenomena. In this paper we propose a novel gene regulatory model based on threshold logic. The approach is developed by a combination of threshold logic properties and perceptron learning techniques. This work does not focus on determination of the pair-wise interactions among genes. Instead, the objective of this work is to generate a model that will describe and predict phenomena associated with a biological system. The utility of the approach is demonstrated by modeling a cellular system of 50 genes. The model could effectively replicate both the steady state and the transient behavior of genes.

Prediction of pairwise gene interaction using threshold logic.

The two important problems of computational biology are the modeling of gene regulatory networks and the study of the network structure of complex biological systems. There is an increased use of mathematical and computational theory techniques to solve both these problems. The Boolean circuit model is one of the most popular modeling paradigms used to model gene regulatory networks. In this paper we try to make use of the properties of threshold logic (an alternative to Boolean logic to design digital circuits) to determine the network structure of gene systems. This approach uses the gene-expression data from microarray experiments as input. The proposed method was first used to build the gene network for a set of genes, proteins, and other molecular components based on in silico data. Then, the method was applied to a biological dataset to build the gene regulatory network for a core set of genes associated with melanoma. Some of the interactions found could be verified by earlier biological experiments reported in published literature. Other interactions that could not be validated by existing biological knowledge can provide insights into the investigation of bio-chemical pathways associated with melanoma development.