gene.neuralcortex.com

neuralcortex.com

neuralcortex

http://www.neuralcortex.com/a_theory.htm

Definition of the Cortex. NC is an indexing system for ultra fast pattern recognition. But, indexing and recognition are often philosophically worlds apart in the theory of machine learning. The question is: how pattern recognition can be achieved by indexing?

neuralcortex.com

neuralcortex

http://www.neuralcortex.com/a_products.htm

Systems designed for NC classification and recognition use highly parallel processing. To take advantage of NC's parallel nature. Small PC-form systems use AMD's new GCN technology to achieve breathtaking performace is a small space. 1M vectors x any 256. Out of 16K features, 1. 1M vectors x any 256 features out of 16K features, 2. 1M vectors x any 256 features out of 16K features, 3. 100M vectors x 16K features (no GCN cores). Neural Cortex technology used for classification and recognition is easily im...

neuralcortex.com

neuralcortex

http://www.neuralcortex.com/a_advantages.htm

See Benchmarks, Speed Comparison. A single NC module supports both mobile applications and large-scale applications (1M vectors x 256 features out. Of 16K features). Multiple NC modules support very large-scale applications (100M and beyond). See Benchmarks, Accuracy.

neuralcortex.com

neuralcortex

http://www.neuralcortex.com/a_benchmarks.htm

Speed Comparison of some GPU-based Classifiers. The following table compares the speed performance of three classifiers, all of which use GPU accelleration. The data set being tested is the 581012 vector FOREST cover data set with 45 features. The first of the classifiers is using a Neural Cortex. NC versus fastest MLcomp classifiers. He neural cortex classifier's. NC accuracy versus other common reference databases.

neuralcortex.com

neuralcortex

http://www.neuralcortex.com/a_generalapp.htm

Application areas for NC include. Automotive (recognition of moving, continuously changing objects). Consumer Electronics and Toys (gesture recognition). Robotics (wireless, off-board intelligence). Scientific Research (large-energy physics, astronomy, discovery of DNA-patterns). Information Technologies (recommenders, search and diagnostic engines, etc.).

neuralcortex.com

neuralcortex

http://www.neuralcortex.com/a_digitalreal.htm

Digital and Real Cortex. Tsunoda, K., Yamane, Y., Nishizaki, M., and Tanifuji, M.: "Complex objects are represented in macaque inferotemporal cortex by the combination of feature columns." Nature Neuroscience. 4, 832-838 (2001). Active pixels of a visual area using a digital neural cortex. Notice in both the real and digital models, several feature regions are covered by multiple colors and others are not. The right side picture above shows. Active pixels of a visual area of digital NC.

neuralcortex.com

neuralcortex

http://www.neuralcortex.com/a_faq.htm

Q Is NC a completely novel learning framework or is it an indexing system, which is distinct from learning. What's the difference between classification and recognition? According to: Pattern Recognition, 3rd edition by S. Theodoridis, K. Koutroumbas, Academic Press. Pattern Recognition is the scientific discipline whose goal is the classification of objects into a number of categories or classes. The picture illustrating this statement can be found in http:/ neuralcortex.com. Can GeneStorm/Neural cortex...

neuralcortex.com

neuralcortex

http://www.neuralcortex.com/index.htm

NC) is a unique multidimensional indexing system for classification, clustering, and. Discovery that achieves new levels of performance in. Neural and cognitive computation. NC outperforms software and hardware based neural networks in terms of speed (. With no compromise in accuracy. Its novel structure makes large-scale. Applications possible on a. Of species by evolving additional cortical columns while keeping their heights at bay. Unlike the above discussed approaches that operate in the N-dimension...

neuralcortex.com

neuralcortex

http://www.neuralcortex.com/a_onlinetest.htm

Data files must be ASCII format with CRLF or LF terminated lines (Windows or Unix). Files may be space or comma separated CSV format, or MLcomp format. (see MLcomp.org/faq/domains. Filenames cannot have spaces. Filenames can use any extension including no extension. CSV files represent fixed length vectors where each feature's ID equals its position in the vector. MLcomp files represent sparse data where each feature is assigned an ID and value separated by a ':'. CSV numeric vector format. If a GPU serv...