DynapSIM: A Fast, Optimizable, and Mismatch Aware Mixed-Signal Neuromorphic Chip Simulator

Download

thesis.pdf

Date

2022-8-22

Author

Çakal, Uğurcan

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

432
views

438
downloads

Mixed-signal neuromorphic processors utilize analog signal processing and digital asynchronous communication, inspired by biological nervous systems' operation principles. Although these architectures provide an enormous power efficiency advantage over existing neural network inference systems, the difficulties in the configuration are one of the most fundamental obstacles in front of developing applications. Limited controllability over the analog hardware parameters, unintended variations inherent to a device's hardware makeup, and linearly inseparable bias space makes this hard to deliver an application that works as intended. It usually requires months of manual calibration effort of highly qualified researchers. Filling the gap, this study presents a software toolchain that allows an offline optimization of a hardware configuration that reflects a spiking neural network implementation. The results show how an abstract spiking neural network accurately and reliably translates into VLSI neuron and synapse configuration in a noisy environment. Proposed methods can be tailored to any mixed-signal neuromorphic processor architecture.

Subject Keywords

Mixed-Signal, Chip Simulation, Dynap-SE1, Dynap-SE2, Neuromorphic Computing, Neuromorphic Hardware, Non-Von-Neumann Computing, Silicon Brain, Spiking Neural Networks

URI

https://hdl.handle.net/11511/98616

Collections

Graduate School of Natural and Applied Sciences, Thesis

Suggestions

OpenMETU
Core

Cooperative relaying schemes for narrow-band frequency hopping wireless ad hoc networks Yenihayat, Güven; Onat, Furuzan Atay; Kolaǧasioǧlu, Ertuǧrul; Yılmaz, Ali Özgür (2011-08-29) Cooperative relaying is a promising technique for improving the performance of weak communication links in wireless networks by exploiting spatial diversity induced by relay transmissions. This paper considers cooperative relaying schemes for a narrow- band frequency hopping system. A novel hybrid relaying scheme called opportunistic hybrid relaying, which is specifically designed for this system, is presented. Unlike most relaying schemes in the literature, opportunistic hybrid relaying uses bandwidth effi...
MIMO Communication Systems Using Hierarchical Modulation Ugur, Yigit; Yılmaz, Ali Özgür (2014-04-25) This paper considers multiple-input multiple-output (MIMO) communication systems using hierarchical modulation. The disadvantage of the maximum-likelihood (ML) receiver is that the computational complexity increases exponentially with the number of transmit antennas and the number of bits per modulation symbol. Using MIMO communication system with 16-ary hierarchical quadrature amplitude modulation (16-HQAM), we propose a receiver structure which can be an alternative to maximum-likelihood (ML) receiver. 16...
Mesh Learning for Object Classification using fMRI Measurements Ekmekci, Ömer; Ozay, Mete; Oztekin, Ilke; GİLLAM, İLKE; Oztekin, Uygar (2013-09-18) Machine learning algorithms have been widely used as reliable methods for modeling and classifying cognitive processes using functional Magnetic Resonance Imaging (fMRI) data. In this study, we aim to classify fMRI measurements recorded during an object recognition experiment. Previous studies focus on Multi Voxel Pattern Analysis (MVPA) which feeds a set of active voxels in a concatenated vector form to a machine learning algorithm to train and classify the cognitive processes. In most of the MVPA methods,...
Measurement correction of a set of analog sun sensors via neural network Sozen, Semsettin Numan; Gokce, Murat; Yavuzyilmaz, Cagatay; Gulmammadov, Farid; Söken, Halil Ersin (2021-06-23) A Neural Network (NN) based method to improve the accuracy of a set of analog Sun sensors is presented. Analog Sun Sensors are commonly used on satellites due to their reduced cost, small size and low power consumption. However, especially in Earth imaging satellites, they are prone to the Earth albedo effects. Magnitude and direction of albedo change depending on the reflection characteristics of the Earth's surface, position and attitude of the satellite and position of the Sun. The albedo may deteriorate...
A PROPOSED APPLICATION OF AN INDUSTRIAL DOS COMPUTER FOR DISTRIBUTION SUBSTATION MONITORING AND CONTROL BATUR, I; Güven, Ali Nezih; OZAY, N (1994-04-01) This paper presents the design and implementation of an intelligent controller intended for use in distribution substations to perform monitoring and control duties within the framework of distribution automation needs. The hardware of the developed controller is a DOS based industrial computer running on ECB-BUS. The application software handles all input/output tasks, data collection, manipulation and control decisions with its flexible structure. This embedded controller measures bus voltage, transform...

Citation Formats

U. Çakal, “DynapSIM: A Fast, Optimizable, and Mismatch Aware Mixed-Signal Neuromorphic Chip Simulator,” M.S. - Master of Science, Middle East Technical University, 2022.