The potential in using backpropagation neural networks for facial verification systems
Article Abstract:
A survey was conducted testing the aptitude of neural networks to verify human faces. The motivation for such a study is the possibility of designing security systems. Pictures of 511 subjects were collected. The pictures captured both profiles and many natural expressions of the subject. Some of the subjects were wearing glasses, sunglasses or hats in some of the pictures. The images were compressed by a magnitude of 100, and converted into image vectors of 1400 pixels. The image vectors were fed into a backpropagation neural network with one hidden layer and one output node. The networks were trained to yield an output that approached 1.0 for the target subject and 0.0 for the other subjects. Neural networks for 37 target subjects were trained using eight different training sets that consisted of different subsets of the data. The networks were then tested on the rest of the data that consisted of 7,000 or more unseen pictures. Our results indicate that a false acceptance rate of less than 1% can be obtained, and a high verification rate can be obtained when certain restrictions are followed. (Reprinted by permission of the publisher.)
Publication Name: SIMULATION
Subject: Engineering and manufacturing industries
ISSN: 0037-5497
Year: 1992
User Contributions:
Comment about this article or add new information about this topic:
Interactive simulation of backpropagation and creeping-random-search learning in neural networks
Article Abstract:
A new environment for interactive neural-network experiments on personal computers, DESIRE/NEUNET, combines a readable matrix language with very fast 'direct execution.' Experimenters enter and screen-edit simulation programs and run them immediately, without the usual annoying delays for compilation and linking. Simulations still run much faster than Microsoft FORTRAN. Users can compose their own help screens and interactive menus. DESIRE/NEUNET permits conventional dynamic-system simulation (up to 1000 differential equations) as well as, or in conjunction with, neural-net simulation and has been used to simulate Hopfield and bidirectional associative memories, transversal predictors, and various competitive-learning schemes. This report exhibits the classical XOR training experiment with backpropagation and creeping-random-search learning. (Reprinted by permission of the publisher.)
Publication Name: SIMULATION
Subject: Engineering and manufacturing industries
ISSN: 0037-5497
Year: 1990
User Contributions:
Comment about this article or add new information about this topic:
A new software technique for sub-model invocation
Article Abstract:
Direct-executing ENHANCED DESIRE simulation on PC-XT-AT and compatible personal computers employs a new method for declaring and invoking sub-model macros and user-defined functions in the DYNAMIC program segment representing the simulation model. Submodel and function invocations are not true macros but compiler procedures; like all of the DYNAMIC segment, they compile into fast in-line code, and no extra translator pass is needed. (Reprinted by permission of the publisher.)
Publication Name: SIMULATION
Subject: Engineering and manufacturing industries
ISSN: 0037-5497
Year: 1987
User Contributions:
Comment about this article or add new information about this topic:
- Abstracts: Real-time simulation and animation for dynamic control systems. Procedural animation blocks in discrete simulation
- Abstracts: A framework for distributed VLSI simulation on a network of workstations
- Abstracts: Toward a unified theory of managing large networks. From personal appliance to personal agent. Accelerators speed logic and fault simulation
- Abstracts: Tips on setting up your workstation. ICs: the brains of a workstation. It looks like a pc--or is it a workstation?
- Abstracts: Biomedical engineering R&D picks up. Expert opinion: imaging and robotics are changing the face of surgery. Gregory Markhov: in love with the light