Intelligent Control of Seizure-Like Activity in a Memristive Neuromorphic Circuit Based on the Hodgkin–Huxley Model

Round 1

Reviewer 1 Report

In this manuscript, the authors reported an intelligent controller for suppressing the seizure-like events in a memristive circuit based on the Hodgkin-Huxley equations. an adaptive neural network is adopted within a Lyapunov-based nonlinear control scheme to attenuate bursting dynamics in the circuit, while compensating for modeling uncertainties and external disturbances. The results confirm the effectiveness of the proposed intelligent controller. However, there are some issues should be addressed be the publication of this manuscript.

1.     What specific application example would be influenced by the seizure-like activity of memristive circuit based on the Hodgkin-Huxley equations.

2.     The author should present the performance comparison of specific application, which based on the conventional controller and this novel controller.

3.     In the reviewer’s opinion, the seizure-like activity of memristive circuit based on the Hodgkin-Huxley equations is similar to the tonic bursting of bio-neuron. Hence, suppressing this activity may influence the reliability for Hodgkin-Huxley equations to emulated bio-neuron functions.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Authors present a neuronal network based control system in order to avoid seizure in Hodgkin-Huxley neurons. The paper is well written and the results interesing and with potential application. So, I recommend publication after considering some issues:

* Maybe the paper tittle is too general. The study is focused on a very specific neuromorphic circuit (HH circuits built with memristors). Please, change the tittle or explain how it could be extended the control proposal to other neuromorphic circuits.

* Authors define and use very specific memristors, whose conductances are given by equations (4) and (5). Why are these expressions used? Are they commonly used? Please, explain the origin of equatiosn (4) and (5) or cite other works. What is the meaning of the state variables x1, x2 and x3? What are their dimensions?

* The same for equations (6)-(8). Why have authors used these expressions?

* More important, how would the study and control results be affected if other memristor classes were employed? That is, if memristors with different conductance dependencies (equations 4-5) and different rate differential equations for the evolution of the state variables (equations 6-8) were used, would the author's control proposal give similar results? How would the results be affected by using different memristors or real memristors (such as those based on resistive switching)?

* Page 5, lines 109-110. Low amplitude in relation to ...what? I think that 28mA is not a low amplitude signal in low power circuits and, of course, in biological circuits.

* Page 6, line 137. I don´t understand the sentence "but also the neglected dynamics of internal states, Equations (6)–(8)." What does it mean? Does it mean that the state variables are supposed to be constant during the control? Please, clarify this issue.

* Please, describe the software implementation (C, matlab, ....) a little bit (section 4).

* For the sake of comparison, authors use a conventional nonlinear schema (page 8, line 161).  What is a "conventional" nonlinear schema? Please, describe it and provide details about its implementation.

* Finally, how would a real or hardware implementation be? That is, what would the main challenges to overcome in a real (not simulated) implementation be? For example, delays in the control system (in the calculation of the NN, for example) could affect the performance of the control system?

* Other minor issues:

- Please, show Iext (equation 3) in figure 2.

- In page 3, line between equations (5) and (6). Equations (5)-(8) define the systems dynamics, not the states themselves. Please, re-write this sentence.

- Figure 2 caption: it should be indicated the model and parameters used for the memristors or provide a reference to the text or table where they are indicated.

- Page 6, line 136. Include xt as part of Iext subscript.

- Page 6, line 142. Figure 5 instead of Figure 4.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

The manuscript (jlpea-1889577), Intelligent Control of Seizure-Like Activity in Memristive Neuromorphic Circuits, shows a very interesting result proposed in intelligent Controller for Memristive Neuromorphic Circuits. Several questions or comments would like to provide here and hope it can improve the manuscript - 

1. For the Intelligent control scheme, how does the accuracy impact or improve as compared to conventional control in memristive neural network computation? 

2. Does the proposed intelligent control scheme here be able to fix the various amplitudes (high vs. low) and speed (fast vs. slow) on memristive circuits?

3. What would be the actual intelligent controller circuit in this proposal and corresponding area, power consumption as compared conventional controller?

4. Can authors provide a simple benchmark Table to compare various controller schemes in recent years for Memristive Neuromorphic Circuits? What is their area, power consumption, speed, performance as compared to current work? This would be quite helpful for this referee and potential readers to get further insight of the research impact and research contribution in this work as compared to other works. 

Due to the above comments, this referee would like to put the manuscript status as "Major Revision" in the current phase. 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors have addressed all of my concerns.

Reviewer 3 Report

Authors have replied to this referee in detail. No further comments provided from this referee.