When approaching a novel visual recognition problem in a specialized image domain, a common strategy is to start with a pre-trained deep neural network and fine-tune it to the specialized domain. If the target domain covers a smaller visual space than the source domain used for pre-training (e.g. ImageNet), the fine-tuned network is likely to be overparameterized. However, applying network prun...

Dynamic (also known as instance-based) ensemble pruning, selects a (potentially) different subset of models from an ensemble during prediction based on the given unknown instance with the goal of maximizing prediction accuracy. This paper models dynamic ensemble pruning as a multi-label classification task, by considering the members of the ensemble as labels. Multi-label training examples are ...

IN ARKANSAS, 'Concord' grapevines trained to the Geneva Double Curtain system have consistently produced larger yields than vines trained to the bilateral cordon or Umbrella Kniffen systems [Ark. Farm Res. 28(5):12; 1978]. However, information was needed on the influence of pruning severity, cane length and shoot positioning on yield and juice quality as determined by percentage of soluble soli...

| Ensembles of multi-layer network is set up to predict the carbon-13 nuclear magnetic resonance (C13 NMR) chemical shifts of a series of mono-substituted benzenes. The descriptors (inputs) used are twelve structural-based vectors that correspond to the calculated H uckel and Gasteiger electron densities of the mono-substituted aromatic systems and four graphical descriptors that correspond to ...

Induction of decision trees is one of the most successful approaches to supervised machine learning. Branching programs are a generalization of decision trees and, by the boosting analysis, exponentially more eeciently learnable than decision trees. In experiments this advantage has not been seen to materialize. Decision trees are easy to simplify using pruning. For branching programs no such a...

Recently, recursive least square (RLS), or extended Kalman ltering (EKF), based algorithms have been demonstrated to be a class of eeective online training methods for neural networks. This paper discusses several aspects of pruning a neural network trained by the RLS based approach. Based on our study, the RLS approach is implicitly a weight decay training algorithm. Also, we derive two prunin...

Pruning provides an important tool for control of nondeterminism in Prolog systems. Current Tabled Prolog systems improve Prolog’s evaluation strategy in several ways, but lack satisfactory support for pruning operations. In this paper we present an extension to the evaluation mechanism of Tabled Prolog to support pruning. This extension builds on the concept of demand to select tables to prune...

An algorithm is proposed to prune the prototype vectors (prototype selection) used in a nearest neighbor classifier so that a compact classifier can be obtained with similar or even better performance. The pruning procedure is error based; a prototype will be pruned if its deletion leads to the smallest classification error increase. Also each pruning iteration is followed by one epoch of Learn...

Abstract Convolutional neural networks (CNNs) have shown good performance in many practical applications. However, their high computational and storage requirements make them difficult to deploy on resource-constrained devices. To address this issue, paper, we propose a novel iterative structured pruning algorithm for CNNs based the recursive least squares (RLS) optimization. Our combines inver...

In this work, pruning techniques for the AdaBoost classifier are evaluated specially aimed for a continuous learning framework in sensors mining applications. To assess the methods, three pruning schemes are evaluated using standard machine-learning benchmark datasets, simulated drifting datasets and real cases. Early results obtained show that pruning methodologies approach and sometimes out-p...