Linear Regression on a Set of Selected Templates from a Pool of Randomly Generated Templates
We study linear regression for two datasets. For the MNIST dataset we do so using max convolutions, whose parameters are generated directly from training images for the digit recognition problem, hence we call them max convolution templates. From a large pool of randomly generated convolutional templates, we select by iterative process the ones which improve defined linear regression minimization problem the most. With these templates, we use linear and logistic regression and achieve high accuracy, comparable with deep neural networks. We explain why, in a production environment, using this approach has advantages over the use of deep neural networks. On a second dataset ‘Adult Data Set’ of income predictions, we show a similar convolution type approach for generating a pool of random templates and show that the same template selection process and linear regression can be used as for the MNIST dataset.
Code (1)
Tasks
regressionMethods 이 논문이 사용한 방법론
Similar Papers 제목 키워드 기반
Offline Handwritten Chinese Text Recognition with Convolutional Neural Networks
Deep learning based methods have been dominating the text recognition tasks in different and multilingual scenarios. The offline handwritten Chinese text recognition (HCTR) is one of the most challenging tasks because it…
Handwritten Chinese Text RecognitionLanguage ModelingLanguage ModellingAre Direct Links Necessary in RVFL NNs for Regression?
A random vector functional link network (RVFL) is widely used as a universal approximator for classification and regression problems. The big advantage of RVFL is fast training without backpropagation. This is because th…
regressionEasyConvPooling: Random Pooling with Easy Convolution for Accelerating Training and Testing
Convolution operations dominate the overall execution time of Convolutional Neural Networks (CNNs). This paper proposes an easy yet efficient technique for both Convolutional Neural Network training and testing. The conv…
GDP Forecasting using Payments Transaction Data
UK GDP data is published with a lag time of more than a month and it is often adjusted for prior periods. This paper contemplates breaking away from the historic GDP measure to a more dynamic method using Bank Account, C…
Self-Adaptable Templates for Feature Coding
Hierarchical feed-forward networks have been successfully applied in object recognition. At each level of the hierarchy, features are extracted and encoded, followed by a pooling step. Within this processing pipeline, th…
image-classificationImage ClassificationObject RecognitionQuantization