Constructive Neural Network Algorithms That Solve Highly Non-separable Problems
نویسندگان
چکیده
Learning from data with complex non-local relations and multimodal class distribution for widely used classification algorithms is still very hard. Even if accurate solution is found the resulting model may be too complex for a given data and will not generalize well. New types of learning algorithms are needed to extend capabilities of standard machine learning systems. Projection pursuit methods can avoid “curse of dimensionality” by discovering interesting structures in lowdimensional subspace. This paper introduces constructive neural architectures based on projection pursuit techniques that are able to discover simplest models of data with inherent highly complex logical structures.
منابع مشابه
Learning Highly Non-separable Boolean Functions Using Constructive Feedforward Neural Network
Learning problems with inherent non-separable Boolean logic is still a challenge that has not been addressed by neural or kernel classifiers. The k-separability concept introduced recently allows for characterization of complexity of non-separable learning problems. A simple constructive feedforward network that uses a modified form of the error function and a window-like functions to localize ...
متن کاملA Comparison of Methods for Learning of Highly Non-separable Problems
Learning in cases that are almost linearly separable is easy, but for highly non-separable problems all standard machine learning methods fail. Many strategies to build adaptive systems are based on the “divide-and-conquer” principle. Constructive neural network architectures with novel training methods allow to overcome some drawbacks of standard backpropagation MLP networks. They are able to ...
متن کاملConstructive Neural Network Learning Algorithms for Multi-category Pattern Classiication Constructive Neural Network Learning Algorithms for Multi-category Pattern Classiication
Constructive learning algorithms ooer an approach for incremental construction of potentially near-minimal neural network architectures for pattern classiication tasks. Such algorithms help overcome the need for ad-hoc and often inappropriate choice of network topology in the use of algorithms that search for a suitable weight setting in an otherwise a-priori xed network architecture. Several s...
متن کاملConstructive Neural Network Learning Algorithms for Multi-category Pattern Classiication
Constructive learning algorithms o er an approach for incremental construction of potentially near-minimal neural network architectures for pattern classi cation tasks. Such algorithms help overcome the need for ad-hoc and often inappropriate choice of network topology in the use of algorithms that search for a suitable weight setting in an otherwise a-priori xed network architecture. Several s...
متن کاملA Topologically Distributed Encoding to Facilitate Learning
We describe how to solve linear-non-separable problems using simple feed-forward perceptrons without hidden layers. A biologically motivated topologically distributed encoding for input data is used. We point out advantages of neural networks compared to classic mathematical algorithms without loosing performance. The Iris-dataset from Fisher 1] is analyzed as a practical example.
متن کامل