Implementation of fast ICA using memristor crossbar arrays for blind image source separations

Abstract

Independent component analysis (ICA) is an unsupervised learning approach for computing the independent components (ICs) from the multivariate signals or data matrix. The ICs are evaluated based on the multiplication of the weight matrix with the multivariate data matrix. This study proposes a novel Pt/Cu:ZnO/Nb:STO memristor crossbar array for the implementation of both ACY ICA and Fast ICA for blind source separation. The data input was applied in the form of pulse width modulated voltages to the crossbar array and the weight of the implemented neural network is stored in the memristor. The output charges from the memristor columns are used to calculate the weight update, which is executed through the voltages kept higher than the memristor Set/Reset voltages (±1.30 V). In order to demonstrate its potential application, the proposed memristor crossbar arrays based fast ICA architecture is employed for image source separation problem. The experimental results demonstrate that the proposed approach is very effective to separate image sources, and also the contrast of the images are improved with an improvement factor in terms of percentage of structural similarity as 67.27% when compared with the software-based implementation of conventional ACY ICA and Fast ICA algorithms.

Gradient Update using Layer-based Importance Ratio in Deep Neural Networks

“…This approach is akin to a ground-up rebuild of sorts. For instance, [7], [9], [10] make use of resistive RAM technology and components such as memristors to mimic the learning capability of a synapse. In other words, through a variety of signal processing techniques, it has been shown that it is possible to emulate properties such as Spike Timing Dependant Plasticity (STDP), thereby creating a circuit which can act as an artificial synapse.…”

“…Furthermore, while certain papers have focused on improving the architecture required to run deep learning algorithms, others have focused on introducing new elements to the circuit which can take advantage of the general learning capability of a neuron. For instance, [7]- [9] make use of memristors to emulate the learning capability of a synapse. [10] uses different modulation techniques to program the learning capability of the memristor.…”

CW_ICA: an efficient dimensionality determination method for independent component analysis

“…ICA aims to determine both the mixing matrix and the source signal matrix. According to the number of observed signals ( p ) and source signals ( q ), ICA methods can be divided into two cases: (over)determined ICA (i.e., ) (e.g., FastICA 11 , JADE 12 , Infomax 13 , etc. ), and underdetermined ICA ( ) (e.g., FastFCA 14 , MAICA 15 , OICD 16 , etc.).…”

A robust AI Agent-based approach to tackle and prevent Money Laundering

“…Radial basis function neural network has been used in [20]. On the other hand, the decision tree approach has been applied in [21]- [24]. In another recent work [25], an adaptive neuro-fuzzy inference system is adopted for the AML problem.…”

“…In another recent work [25], an adaptive neuro-fuzzy inference system is adopted for the AML problem. Among supervised learning works in the AML domain, [19], [20] have adopted transaction features such as sum and frequency of monetary transactions and [21], [24] have developed models with CRM features. However, there is no example combining these two characteristics.…”