Cooling schedules for optimal annealing
Mathematics of Operations Research
Selecting typical instances in instance-based learning
ML92 Proceedings of the ninth international workshop on Machine learning
Machine Learning - Special issue on learning with probabilistic representations
Local Search in Combinatorial Optimization
Local Search in Combinatorial Optimization
Combining the Perceptron Algorithm with Logarithmic Simulated Annealing
Neural Processing Letters
Facts, Conjectures, and Improvements for Simulated Annealing
Facts, Conjectures, and Improvements for Simulated Annealing
ICANN '02 Proceedings of the International Conference on Artificial Neural Networks
Bounded-depth threshold circuits for computer-assisted CT image classification
Artificial Intelligence in Medicine
Fuzzy CoCo: a cooperative-coevolutionary approach to fuzzy modeling
IEEE Transactions on Fuzzy Systems
Logarithmic simulated annealing for X-ray diagnosis
Artificial Intelligence in Medicine
Neural-network feature selector
IEEE Transactions on Neural Networks
Hi-index | 0.00 |
The LSA machine is an effective method for predicting a class from linear separable data. LSA machine is based on the combination of Logarithmic Simulated Annealing with the Perceptron Algorithm. In this paper we present and compare the classification accuracy of the LSA machine on two medical databases a) the Winsconsin Breast Cancer Database, which is a binary database with two associated classes and b) the Diabetic Patient Management Database, which is a multicategory database with four associated classes. Many researchers use the Winsconsin Breast Cancer Database (WBCD) database, as a benchmark database for testing their systems. The WBCD database consists of 699 samples with 9 input attributes, The LSA machine is trained on 50% and 75% of the entire dataset and in both cases we obtain a classification accuracy of 98.8% on the remaining samples. This classification accuracy on the test set of samples, to our best of knowledge is the highest reported in the literature. The Diabetic Patient Management database consists of 746 samples with 18 input values and an associated class label denoting one of the four treatments for the patient. The LSA machine for comparison reasons is trained on the 646 samples of the database, obtaining stable classification accuracy over 74% for all four classes, with highest classification accuracy of 87%.