Skimming Digits: Neuromorphic Classification of Spike-Encoded Images
The growing demands placed upon the field of computer vision has renewed the focus on alternative visual scene representations and processing paradigms. Silicon retinea provide an alternative means of imaging the visual environment, and produce frame-free spatio-temporal data. This paper presents an...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2016-04-01
|
| Series: | Frontiers in Neuroscience |
| Subjects: | |
| Online Access: | http://journal.frontiersin.org/Journal/10.3389/fnins.2016.00184/full |
| _version_ | 1818931700306018304 |
|---|---|
| author | Gregory Kevin Cohen Gregory Kevin Cohen Garrick eOrchard Sio Hoi eIeng Jonathan eTapson Ryad Benjamin Benosman André evan Schaik |
| author_facet | Gregory Kevin Cohen Gregory Kevin Cohen Garrick eOrchard Sio Hoi eIeng Jonathan eTapson Ryad Benjamin Benosman André evan Schaik |
| author_sort | Gregory Kevin Cohen |
| collection | DOAJ |
| description | The growing demands placed upon the field of computer vision has renewed the focus on alternative visual scene representations and processing paradigms. Silicon retinea provide an alternative means of imaging the visual environment, and produce frame-free spatio-temporal data. This paper presents an investigation into event-based digit classification using N-MNIST,a neuromorphic dataset created with a silicon retina, and the Synaptic Kernel Inverse Method (SKIM), a learning method based on principles of dendritic computation. As this work represents the first large-scale and multi-class classification task performed using the SKIM network, it explores different training patterns and output determination methods necessary to extend the original SKIM method to support multi-class problems. Making use of SKIM networks applied to real-world datasets, implementing the largest hidden layer sizes and simultaneously training the largest number of output neurons, the classification system achieved a best-case accuracy of 92.87% for a network containing 10,000 hidden layer neurons. These results represent the highest accuracies achieved against the dataset to date and serves to validate the application of the SKIM method to event-based visual classification tasks. Additionally, the study found that using a square pulse as the supervisory training signal produced the highest accuracy for most output determination methods, but the results also demonstrate that an exponential pattern is better suited to hardware implementations as it makes use of the simplest output determination method based on the maximum value. |
| first_indexed | 2024-12-20T04:20:45Z |
| format | Article |
| id | doaj.art-99b121db1e5b4e6a9fa7273bc9738d49 |
| institution | Directory Open Access Journal |
| issn | 1662-453X |
| language | English |
| last_indexed | 2024-12-20T04:20:45Z |
| publishDate | 2016-04-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Neuroscience |
| spelling | doaj.art-99b121db1e5b4e6a9fa7273bc9738d492022-12-21T19:53:39ZengFrontiers Media S.A.Frontiers in Neuroscience1662-453X2016-04-011010.3389/fnins.2016.00184191715Skimming Digits: Neuromorphic Classification of Spike-Encoded ImagesGregory Kevin Cohen0Gregory Kevin Cohen1Garrick eOrchard2Sio Hoi eIeng3Jonathan eTapson4Ryad Benjamin Benosman5André evan Schaik6Western Sydney UniversityInstitut de la VisionNational University of SingaporeInstitut de la VisionWestern Sydney UniversityInstitut de la VisionWestern Sydney UniversityThe growing demands placed upon the field of computer vision has renewed the focus on alternative visual scene representations and processing paradigms. Silicon retinea provide an alternative means of imaging the visual environment, and produce frame-free spatio-temporal data. This paper presents an investigation into event-based digit classification using N-MNIST,a neuromorphic dataset created with a silicon retina, and the Synaptic Kernel Inverse Method (SKIM), a learning method based on principles of dendritic computation. As this work represents the first large-scale and multi-class classification task performed using the SKIM network, it explores different training patterns and output determination methods necessary to extend the original SKIM method to support multi-class problems. Making use of SKIM networks applied to real-world datasets, implementing the largest hidden layer sizes and simultaneously training the largest number of output neurons, the classification system achieved a best-case accuracy of 92.87% for a network containing 10,000 hidden layer neurons. These results represent the highest accuracies achieved against the dataset to date and serves to validate the application of the SKIM method to event-based visual classification tasks. Additionally, the study found that using a square pulse as the supervisory training signal produced the highest accuracy for most output determination methods, but the results also demonstrate that an exponential pattern is better suited to hardware implementations as it makes use of the simplest output determination method based on the maximum value.http://journal.frontiersin.org/Journal/10.3389/fnins.2016.00184/fullOpiumobject classificationmulti-classSkimN-MNIST |
| spellingShingle | Gregory Kevin Cohen Gregory Kevin Cohen Garrick eOrchard Sio Hoi eIeng Jonathan eTapson Ryad Benjamin Benosman André evan Schaik Skimming Digits: Neuromorphic Classification of Spike-Encoded Images Frontiers in Neuroscience Opium object classification multi-class Skim N-MNIST |
| title | Skimming Digits: Neuromorphic Classification of Spike-Encoded Images |
| title_full | Skimming Digits: Neuromorphic Classification of Spike-Encoded Images |
| title_fullStr | Skimming Digits: Neuromorphic Classification of Spike-Encoded Images |
| title_full_unstemmed | Skimming Digits: Neuromorphic Classification of Spike-Encoded Images |
| title_short | Skimming Digits: Neuromorphic Classification of Spike-Encoded Images |
| title_sort | skimming digits neuromorphic classification of spike encoded images |
| topic | Opium object classification multi-class Skim N-MNIST |
| url | http://journal.frontiersin.org/Journal/10.3389/fnins.2016.00184/full |
| work_keys_str_mv | AT gregorykevincohen skimmingdigitsneuromorphicclassificationofspikeencodedimages AT gregorykevincohen skimmingdigitsneuromorphicclassificationofspikeencodedimages AT garrickeorchard skimmingdigitsneuromorphicclassificationofspikeencodedimages AT siohoieieng skimmingdigitsneuromorphicclassificationofspikeencodedimages AT jonathanetapson skimmingdigitsneuromorphicclassificationofspikeencodedimages AT ryadbenjaminbenosman skimmingdigitsneuromorphicclassificationofspikeencodedimages AT andreevanschaik skimmingdigitsneuromorphicclassificationofspikeencodedimages |