Generative versus discriminative training of RBMs for classification of fMRI images
نویسندگان
چکیده
Neuroimaging datasets often have a very large number of voxels and a very small number of training cases, which means that overfitting of models for this data can become a very serious problem. Working with a set of fMRI images from a study on stroke recovery, we consider a classification task for which logistic regression performs poorly, even when L1or L2regularized. We show that much better discrimination can be achieved by fitting a generative model to each separate condition and then seeing which model is most likely to have generated the data. We compare discriminative training of exactly the same set of models, and we also consider convex blends of generative and discriminative training.
منابع مشابه
Static Gesture Recognition with Restricted Boltzmann Machines
In this paper I investigate a new technique for the recognition of static gestures (poses) from laptop camera images. I apply Restricted Boltzmann Machines (RBMs) to model the manifold of 3 human gestures: pointing, thumbs up, fingers spread, as well as the default no-gesture case. The generative RBM model performs significantly better than other classification techniques including classical di...
متن کاملExploiting local structure in Boltzmann machines
Restricted Boltzmann Machines (RBM) are well-studied generative models. For image data, however, standard RBMs are suboptimal, since they do not exploit the local nature of image statistics. We modify RBMs to focus on local structure by restricting visible-hidden interactions. We model longrange dependencies using direct or indirect lateral interaction between hidden variables. While learning i...
متن کاملExploiting local structure in stacked Boltzmann machines
Restricted Boltzmann Machines (RBM) are well-studied generative models. For image data, however, standard RBMs are suboptimal, since they do not exploit the local nature of image statistics. We modify RBMs to focus on local structure by restricting visible-hidden interactions. We model long-range interactions using direct or indirect lateral interaction between hidden variables. While learning ...
متن کاملLearning Algorithms for the Classification Restricted Boltzmann Machine
Recent developments have demonstrated the capacity of restricted Boltzmann machines (RBM) to be powerful generative models, able to extract useful features from input data or construct deep artificial neural networks. In such settings, the RBM only yields a preprocessing or an initialization for some other model, instead of acting as a complete supervised model in its own right. In this paper, ...
متن کاملDiscriminative Restricted Boltzmann Machines are Universal Approximators for Discrete Data
This report proofs that discriminative Restricted Boltzmann Machines (RBMs) are universal approximators for discrete data by adapting existing universal approximation proofs for generative RBMs. Discriminative Restricted Boltzmann Machines are Universal Approximators for Discrete Data Laurens van der Maaten Pattern Recognition & Bioinformatics Laboratory Delft University of Technology
متن کامل