![[PukiWiki]](image/pukiwiki.png "[PukiWiki]")
As artificial neural networks (ANNs) are executed in a fully synchronized manner, most of the computation is inherently uselessly done. For instance, a visual-based ANN performing object tracking analyses the entire image at each timestamp while most of the information remains unchanged within small time intervals. Another consequence of the synchronous execution is that traditional ANNs are not optimized to extract the timing information. In contrast, neuromorphic systems are characterized by the asynchronous and independent execution of the neurons within the neural net, therefore enabling more flexibility, as well as the ability to learn the timing information. [www].
It has been understood that spiking neural networks (SNNs) and synaptic devices can achieve high energy efficiency and fault tolerance in complex tasks by emulating biological neurons and synapses. Hence, they have a big potential for the replacement of traditional von Neumann computers. [WWW.ref].
SNNs can achieve better power efficiency when more powerful encoding methods are applied, or when SNNs are optimized for short latency. Second, the inference results in SNNs can be gotten faster than ANNs, though the precision may be lower. The inference precision would then rise with more evidence accumulated over time Ref. Spiking neurons are only activated when sufficient signals are integrated from other neurons, therefore the neural activities at the network level are usually sparse, which can make executing SNNs by matrix-based operation on a traditional CPU or GPU inefficient.Ref.
One issue in SNNs is that the spike function is non-differentiable, making it impossible to use backpropagation to train the network. To address this issue, several solutions have been proposed, such as converting DNNs to SNNs, and approximating the derivative of the spike function [ref]
Neuromorphic hardware is designed to accelerate the running of SNNs by allocating spiking neurons and synapses to neuromorphic chips and it can increase the power efficiency of SNNs by event-based computing.
The goal of this Doctoral thesis is to research and implement a Deep Spiking Neural Network Architecture with on-chip learning scheme and based on the previously developed schemes NASH1
The expected outputs from this research are:
| Date | Task |
| ☑️April 1st 2019 - August 30, 2019 | Survey, etc. |
| ☑️September 1st, 2019 - July 31,2020 | Complete SNPC on-chip learning with STDP |
| September 21, 2020 | JoS paper draft |
| ☑️August 16, 2020 to September 21, 2020 | Understand and run demo of k-means clustering based fault-tolerant multicast spike routing algorithm |
| ☑️August 16, 2020 to September 21, 2020 | Replace spike injector with SNPC and run again the demo of k-means clustering based fault-tolerant multicast spike routing algorithm |
| ☑️September 1, 2020 to December 4, 2020 | Integrate SNPC into 3D-NoC architecture |
| ☑️September 1, 2020 to February 28, 2021 | Prepare journal paper draft on the 3D-NoC SNN |
| ☑️March 1, 2021 to March 3, 2021 | Prepare journal draft for MDPI Electronics |
| ☑️March 3, 2021 to March 5, 2021 | Prepare conference draft for UNET21 |
| ☑️March 5, 2021 to March 21, 2021 | Prepare conference draft for Reversible Computing 13 |
| ☑️March 19, 2021 to March 24, 2021 | Update journal manuscript for MDPI Electronics |
| ☑️April 2, 2021 to April 17, 2021 | NASH tutorial |
| ☑️April 6, 2021 to April 9, 2021 | Desertation draft structure |
| May 18, 2021 to May 30, 2021 | Investigate more biological benchmarks on NASH, including Hand Gesture Recognition |
| ☑️April 28, 2021 to May 6, 2021 | UNET Camera ready paper |
| ☑️April 18, 2021 to May 21, 2021 | Dissertation draft completion preparation of documents for the Doctoral Dissertation Preliminary Review |
| ☑️May 10, 2021 | Preparation of documents for the Doctoral Dissertation Preliminary Review |
| ☑️June 7 | First Dissertation Preliminary Presentation Rehearsal |
| ☑️June 14 | Second Dissertation Preliminary Presentation Rehearsal |
| July 1, 2021 to September 31, 2021 | Begin work on NeuroSys |
| ☑️July 1, 2021 to July 12, 2021 | Complete initial exploration of RBTC application |
| ☑️August 1, 2021 to August 7, 2021 | Survey on invasive and non-invasive prosthetics, investigate gesture recognition ANN implementation on Raspberry pi |
| August 8, 2021 to August 14, 2021 | Update prosthetics journal draft with Williams, Begin implementation of gesture recognition ANN on raspberry pi |
| August 15, 2021 to August 28, 2021 | Implementation of gesture recognition ANN on raspberry pi |
| September 1, 2021 to September 31, 2022 | Preparation of Dissertation Draft version 2, SNN implementatioin on raspberry pi |
| ☑️August 26, 2021 to September 3, 2021 | Journal Paper first draft |
| January 3, 2021 | Completion of Final review slides |
| January 6, 2022 | Final review presentation rehearsal |
| December 13, 2021 to December 31, 2021 | Conference paper complete draft |
| Date | Task |
| ☑️March 22, 2021 | Submission of dissertation title and list of referees |
| ☑️May 17, 2021 | Submission of the documents for the Doctoral Dissertation Preliminary Review |
| ☑️June 18, 2021 | Doctoral dissertation Rreliminary Review |
| ☑️December 17, 2021 | Submission of the documents for the Doctoral Dissertation final Review |
| January 12, 2022 | Doctoral Dissertation Final Review |
| January 24, 2022 | Committee Meeting |
| January 28, 2022 | Submission of the materials for the presentation |
| February 15, 2022 | Submission of finalized Dissertations and their abstracts |
| February 15, 2022 | Submission of the Abstracts of the Results Regarding the Final Doctoral Dissertation Review |
| February 15, 2022 | Submission of Consent to Use of Academic Paper / Repository Registration Request Form |
| February 15, 2022 | Submission of the Application Form for the Academic Degree |
| February 16, 2022 | Dissertation presentation |
