Many applications require classification of examples into one of several classes. A common way of designing such classifiers is to determine the class based on the outputs of several binary classifiers. We consider some of the most popular methods for combining the decisions of the binary classifiers, and improve ex-isting bounds on the error rates of the combined classifier over the training set. We also describe a new method for combining binary classifiers. The method is based on stacking a neural network and, when used with support vector machines as the binary learners, substantially decreased the error rate in two vowel classification tasks. 1.

We propose a new architecture for deploying speech recognition over the Internet. The client performs the recognition, but is assisted by the server who computes the speech parameters. To demonstrate the architecture, we developed a Java-based Web-navigation system where the precomputed HMM models of the hyperlinked words are stored on the Web page and downloaded by the client. We tested the system on a digit-recognition example. The results show that with quantization and compression of the speech parameters, good recognition can be achieved in acceptable download and calculation time even on clients with modest connection speeds and computational powers.