Impact of distance-based metric learning on classification and visualization model performance and structure–activity landscapes

Abstract

This study concerns large margin nearest neighbors classifier and its multi-metric extension as the efficient approaches for metric learning which aimed to learn an appropriate distance/similarity function for considered case studies. In recent years, many studies in data mining and pattern recognition have demonstrated that a learned metric can significantly improve the performance in classification, clustering and retrieval tasks. The paper describes application of the metric learning approach to in silico assessment of chemical liabilities. Chemical liabilities, such as adverse effects and toxicity, play a significant role in drug discovery process, in silico assessment of chemical liabilities is an important step aimed to reduce costs and animal testing by complementing or replacing in vitro and in vivo experiments. Here, to our knowledge for the first time, a distance-based metric learning procedures have been applied for in silico assessment of chemical liabilities, the impact of metric learning on structure-activity landscapes and predictive performance of developed models has been analyzed, the learned metric was used in support vector machines. The metric learning results have been illustrated using linear and non-linear data visualization techniques in order to indicate how the change of metrics affected nearest neighbors relations and descriptor space.

Performance of Deep and Shallow Neural Networks, the Universal Approximation Theorem, Activity Cliffs, and QSAR

“…Indeed, very rugged and bumpy activity landscapes with numerous activity cliffs with respect to input descriptors using low‐level representations might appear to be very smooth and simple when transformed to high‐level representations in deep‐learning systems; this is the essence of representation learning. The ability of metric learning, a kind of linear representation learning, to eliminate activity cliffs in activity landscapes has recently been demonstrated . One can expect that non‐linear representation learning provided by neural networks should give an even greater effect.…”

A renaissance of neural networks in drug discovery

“…The ability of metric learning, a kind of linear representation learning, to eliminate activity cliffs in activity landscapes has recently been demonstrated. 110 One can expect that nonlinear representation learning provided by neural networks should give an even greater effect.…”

Core Techniques on Designing Network for Data Centers: A Comparative Analysis

“…The average throughput for the workloads in each architecture is measured by averaging over three independent runs. Later we would like to use the distance metrics [5][6][7] to modify the throughput metrics.…”