Martigny - Valais - Suisse Combining Linear
نویسنده
چکیده
A polychotomizer which assigns the input to one ofK K is constructed using a set of dichotomizers which assign the input to one of two classes We propose techniques to construct a set of linear dichotomizers whose combined decision forms a nonlinear polychotomizer to extract structure from data One way is using error correcting output codes ECOC We propose to incorporate soft weight sharing in training a multilayer perceptron MLP to force the second layer weights to a bimodal distribution to be able to interpret them as the decomposition matrix of classes in terms of dichotomizers This technique can also be used to netune a set of dichotomizers already generated for example using ECOC in such a case ECOC de nes the target values for hidden units in an MLP facilitating training Simulation results on eight datasets indicate that compared with a linear one per class polychotomizer pairwise linear dichotomizers and ECOC based linear dichotomizers this method generates more accurate classi ers We also propose and test a method of incremental construction whereby the required number of dichotomizers is determined automatically as opposed to assumed a priori
منابع مشابه
IDIAP Martigny - Valais - Suisse Fast Object Detection using MLP and FFT
We propose a new technique that speeds up signi cantly the time needed by a trained network MLP in our case to detect a face in a large image We reformulate neural activities in the hidden layer of the MLP in terms of lter convolution enabling the use of Fourier transform for an e cient computation of the neural activities A formal proof and a complexity analysis are presented Finally some exam...
متن کاملMartigny - Valais - Suisse Illumination � robust Pattern Matching using Distorted Histograms Georg
It is argued that global illumination should be modeled separately from other incidents that change the appearance of objects The e ects of intensity variations of the global illumination are discussed and constraints deduced that restrict the shape of a function that maps the histogram of a template to the histogram of an image location This approach is illustrated for simple pattern matching ...
متن کاملIDIAP Martigny - Valais - Suisse Continuous Audio � Visual Speech Recognition
We address the problem of robust lip tracking, visual speech feature extraction, and sensor integration for audiovisual speech recognition applications. An appearance based model of the articulators, which represents linguistically important features, is learned from example images and is used to locate, track, and recover visual speech information. We t a c kle the problem of joint temporal mo...
متن کاملIDIAP Martigny - Valais - Suisse Multi � Modal Data Fusion for Person Authentication
In the context of multi modal person authentication a set of experts face recognizer speaker recognizer etc give their opinion about the identity of an individual The opinions of the experts can be combined to form a nal decision rejecting or accepting the claim We show that the nal decision is a binary classi cation problem and propose to solve it by a Support Vector Machine SVM We compare our...
متن کامل