Online Similarity Learning for Big Data with Overfitting

Abstract

Abstract-In this paper, we propose a general model to address the overfitting problem in online similarity learning for big data, which is generally generated by two kinds of redundancies: 1) feature redundancy, that is there exists redundant (irrelevant) features in the training data; 2) rank redundancy, that is non-redundant (or relevant) features lie in a low rank space. To overcome these, our model is designed to obtain a simple and robust metric matrix through detecting the redundant rows and columns in the metric matrix and constraining the remaining matrix to a low rank space. To reduce feature redundancy, we employ the group sparsity regularization, i.e., the ' 2;1 norm, to encourage a sparse feature set. To address rank redundancy, we adopt the low rank regularization, the max norm, instead of calculating the SVD as in traditional models using the nuclear norm. Therefore, our model can not only generate a low rank metric matrix to avoid overfitting, but also achieves feature selection simultaneously. For model optimization, an online algorithm based on the stochastic proximal method is derived to solve this problem efficiently with the complexity of Oðd 2 Þ. To validate the effectiveness and efficiency of our algorithms, we apply our model to online scene categorization and synthesized data and conduct experiments on various benchmark datasets with comparisons to several state-of-the-art methods. Our model is as efficient as the fastest online similarity learning model OASIS, while performing generally as well as the accurate model OMLLR. Moreover, our model can exclude irrelevant / redundant feature dimension simultaneously.

Incorporating Data Context to Cost-Effectively Automate End-to-End Data Wrangling

“…We will investigate the effect of additional data context types on the wrangling pipeline, and on other wrangling stages such as Web data extraction [49]. To further address time-varying variety and veracity problems in data wrangling, we will investigate feedback-based learning and model refinement techniques such as presented in [42] or [50]. Furthermore, we are exploring how to combine evidence gained from data context with user preferences, as shown in [23], to elaborate the possibilities in tailoring a data product for users with different requirements.…”

The House Price Prediction Using Machine Learning Algorithm: The Case of Jinan, China

“…Feature selection is a process of deleting unrelated or redundant and preserving important features so that features that remain can describe more accurate models. Redundant and unrelated features are noise in machine learning models because they will waste the computational performance of the model and make the model overfit to lose accuracy [11]. There are three feature selection methods: filter, package, and embedded selection.…”

Similarity Based Prediction System using Machine Learning Algorithms in Big Data Analytics

“…al. [1] describes a model for big data to control the overfitting problem which where comes under online similarity learning. The model provides simple and robust metric matrix for finding redundant rows and columns in the metric matrix.…”

“…In the recent research environment, big data [1] is playing a major role to maintain high volumes of data. Many sectors are implemented a big data and analytics like Agricultural, Banking and Online Marketing, etc.…”