A novel adaptive learning algorithm for low-dimensional feature space using memristor-crossbar implementation and on-chip training

Klidbary, Sajad Haghzad; Shouraki, Saeed Bagheri

doi:10.1007/s10489-018-1202-6

Cited by 11 publications

(4 citation statements)

References 58 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The implementation of ink drop spread (IDS) in 2D plans on the memristor-crossbar structure is presented by Merrikh-Bayat et al [27]. There are also other references related to the implementation of the ALM (Active Learning Method) algorithm on memristor-crossbar [28][29][30]. Memristor has also been used as a synapse in neuromorphic computing [31][32][33][34].…”

Section: The Memristor Modelingmentioning

confidence: 99%

See 1 more Smart Citation

Memristor Crossbar-Based Hardware Implementation of Type-2 Fuzzy Membership Function and On-Chip Learning

Klidbary

Javadian

Omidi

et al. 2021

IJE

View full text Add to dashboard Cite

Section: The Memristor Modelingmentioning

confidence: 99%

“…It can be mentioned that by using the ladder structure [28,29], changes can be made as a Gaussian function in the memristance of memristors. As mentioned in section 3, Vertical Slice Representation (Vertical Slice Centroid) for T2F set is considered here.…”

Section: Proposed Hardwarementioning

confidence: 99%

Memristor Crossbar-Based Hardware Implementation of Type-2 Fuzzy Membership Function and On-Chip Learning

Klidbary

Javadian

Omidi

et al. 2021

IJE

View full text Add to dashboard Cite

“…In recent years, the rapid development of electronic components made it possible to simulate neural networks on circuits, and scholars have carried out a variety of explorations: for example, using a memristor-based multiplication circuit to improve the image processing ability of the cellular neural network (CNN); [19] realizing the ink drop spread (IDS) effect on the neural network by a memristor's fuzzy logic properties; [20] simulating the weight update progress of synapses by a memristor-bridge to complete the random weight change (RWC) learning algorithm on multi-layer neu-ral networks. [21] Evidently, all of this reasearch should be attributed to the creation of memristors.…”

Section: Introductionmentioning

confidence: 99%

A spintronic memristive circuit on the optimized RBF-MLP neural network

Gao¹,

Li²,

Jiang³

et al. 2022

Chinese Phys. B

View full text Add to dashboard Cite

Radial basis function network (RBF) has excellent generalization ability and approximation accuracy when parameters set properly.However, relying only on traditional methods,it is tough to obtain optimal network's parameters and construct a stabile model as well.In view of this,RBF-MLP neural network is proposed in this article.In the form of connecting two networks to work cooperatively,the RBF's parameter can be adjusted adaptively by the structure of multi-layer perceptron (MLP) so as to realize the effect of backpropagation updating error. Furthermore, genetic algorithm is used to optimize the network's hidden layer so as to confirm the optimal neurons (basis function) number automatically. In addition, a memristive circuit model is proposed to realize the neural network's operation mentioned above based on the character of spin memristors. It is verified that the network can adaptively construct a network model with outstanding robustness and stably achieve 98.33% accuracy in processing the MNIST dataset classification task. The experimental results show that the method mention above has application value.

show abstract

“…It not only significantly increases the data storage capacity of the devices, but also allows stronger information processing ability in neuromorphic systems. Generally, multiple conductance states of the memristor devices should be achieved in ways that are as simple as possible for practical usage [27][28][29][30]. Nonetheless, most of the studies until now were implemented with complex programming schemes to achieve multi-conductance characteristics, in which the involvement of varying current compliance and voltages leads to a heavy burden on the system circuit design [31][32][33][34].…”

Section: Introductionmentioning

confidence: 99%

Improving the Recognition Accuracy of Memristive Neural Networks via Homogenized Analog Type Conductance Quantization

et al. 2020

View full text Add to dashboard Cite

Conductance quantization (QC) phenomena occurring in metal oxide based memristors demonstrate great potential for high-density data storage through multilevel switching, and analog synaptic weight update for effective training of the artificial neural networks. Continuous, linear and symmetrical modulation of the device conductance is a critical issue in QC behavior of memristors. In this contribution, we employ the scanning probe microscope (SPM) assisted electrode engineering strategy to control the ion migration process to construct single conductive filaments in Pt/HfOx/Pt devices. Upon deliberate tuning and evolution of the filament, 32 half integer quantized conductance states in the 16 G0 to 0.5 G0 range with enhanced distribution uniformity was achieved. Simulation results revealed that the numbers of the available QC states and fluctuation of the conductance at each state play an important role in determining the overall performance of the neural networks. The 32-state QC behavior of the hafnium oxide device shows improved recognition accuracy approaching 90% for handwritten digits, based on analog type operation of the multilayer perception (MLP) neural network.

show abstract

A novel adaptive learning algorithm for low-dimensional feature space using memristor-crossbar implementation and on-chip training

Cited by 11 publications

References 58 publications

Memristor Crossbar-Based Hardware Implementation of Type-2 Fuzzy Membership Function and On-Chip Learning

Memristor Crossbar-Based Hardware Implementation of Type-2 Fuzzy Membership Function and On-Chip Learning

A spintronic memristive circuit on the optimized RBF-MLP neural network

Improving the Recognition Accuracy of Memristive Neural Networks via Homogenized Analog Type Conductance Quantization

Contact Info

Product

Resources

About