Diversity Regularization of Latent Variable Models: Theory, Algorithm and Applications
نویسنده
چکیده
Latent Variable Models (LVMs) are a family of machine learning (ML) models that have been widely used in text mining, computer vision, computational biology, recommender system, to name a few. One central task in machine learning is to extract the latent knowledge and structure from observed data and LVMs elegantly fit into this task. LVMs consist of observed variables used for modeling observed data and latent variables aimed to characterize the hidden knowledge and structure. The interaction between the observed and latent variables encodes modelers’ prior belief regarding how the observed data is generated from or correlated with the latent knowledge and structure. Under LVMs, extracting knowledge from data corresponds to inferring the value of the latent variables given the observed ones.
منابع مشابه
Latent Variable Selection for Multidimensional Item Response Theory Models via L1 Regularization
We develop a latent variable selection method for multidimensional item response theory models. The proposed method identifies latent traits probed by items of a multidimensional test. Its basic strategy is to impose an L1 penalty term to the log-likelihood. The computation is carried out by the expectationmaximization algorithm combined with the coordinate descent algorithm. To the authors’ be...
متن کاملDiversity-Promoting Bayesian Learning of Latent Variable Models
In learning latent variable models (LVMs), it is important to effectively capture infrequent patterns and shrink model size without sacrificing modeling power. Various studies have been done to “diversify” a LVM, which aim to learn a diverse set of latent components in LVMs. Most existing studies fall into a frequentist-style regularization framework, where the components are learned via point ...
متن کاملLearning Latent Space Models with Angular Constraints
The large model capacity of latent space models (LSMs) enables them to achieve great performance on various applications, but meanwhile renders LSMs to be prone to overfitting. Several recent studies investigate a new type of regularization approach, which encourages components in LSMs to be diverse, for the sake of alleviating overfitting. While they have shown promising empirical effectivenes...
متن کاملUsing multivariate generalized linear latent variable models to measure the difference in event count for stranded marine animals
BACKGROUND AND OBJECTIVES: The classification of marine animals as protected species makes data and information on them to be very important. Therefore, this led to the need to retrieve and understand the data on the event counts for stranded marine animals based on location emergence, number of individuals, behavior, and threats to their presence. Whales are g...
متن کاملOn the Generalization Error Bounds of Neural Networks under Diversity-Inducing Mutual Angular Regularization
Recently diversity-inducing regularization methods for latent variable models (LVMs), which encourage the components in LVMs to be diverse, have been studied to address several issues involved in latent variable modeling: (1) how to capture long-tail patterns underlying data; (2) how to reduce model complexity without sacrificing expressivity; (3) how to improve the interpretability of learned ...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 2015