Skip to main content
stat
SMAC Talk - Lixiang Zhang
Add to Calendar 2020-02-14T15:10:00 2020-02-14T16:00:00 UTC SMAC Talk - Lixiang Zhang 327 Thomas Building, University Park, PA
Start DateFri, Feb 14, 2020
10:10 AM
to
End DateFri, Feb 14, 2020
11:00 AM
Presented By
Lixiang Zhang

Deep Variable-Block Chain with Adaptive Variable Selection

Event Series:

Abstract

The architectures of deep neural networks (DNN) rely heavily on the underlying grid structure of variables, for instance, the lattice of pixels in an image. For general high dimensional data with variables not associated with a grid, the multi-layer perceptron and deep brief network are often used. However, it is frequently observed that those networks do not perform competitively and they are not helpful for identifying important variables. In this paper, we propose a new neural network called Deep Variable-Block Chain (DVC) that imposes on variable blocks a chain structure so that DVC can leverage the constructed grid. Because the variable blocks are used for classification in a sequential manner, we further develop the capacity of selecting variables adaptively according to a number of regions trained by a decision tree. Our experiments show that DVC outperforms other generic DNNs and other strong classifiers. Moreover, DVC can achieve high accuracy at much-reduced dimensionality and sometimes reveals drastically different sets of relevant variables for different regions.