Evaluating Regularized Anchor Words
نویسندگان
چکیده
We perform a comprehensive examination of the recently proposed anchor method for topic model inference using topic interpretability and held-out likelihood measures. After measuring the sensitivity to the anchor selection process, we incorporate L2 and Beta regularization into the optimization objective in the recovery step. Preliminary results show that L2 improves heldout likelihood, and Beta regularization improves topic interpretability.
منابع مشابه
Anchors Regularized: Adding Robustness and Extensibility to Scalable Topic-Modeling Algorithms
Spectral methods offer scalable alternatives to Markov chain Monte Carlo and expectation maximization. However, these new methods lack the rich priors associated with probabilistic models. We examine Arora et al.’s anchor words algorithm for topic modeling and develop new, regularized algorithms that not only mathematically resemble Gaussian and Dirichlet priors but also improve the interpretab...
متن کاملA Survey of Currency Anchor Selection in East-West Asia
As the number of independent countries increases and their economies become more integrated, we would expect to observe more multi-country currency :::union:::s. This paper explores the pros and cons for different countries to adopt as an anchor the US Dollar, the Euro or the Yen. In addition, it addresses the question of how co-movement of outputs and prices would respond to the formation of ...
متن کاملEvent Based Emotion Classification for News Articles
Reading of news articles can trigger emotional reactions from its readers. But comparing to other genre of text, news articles that are mainly used to report events, lack emotion linked words and other features for emotion classification. In this paper, we propose an event anchor based method for emotion classification for news articles. Firstly, we build an emotion linked news corpus through c...
متن کاملIs Your Anchor Going Up or Down? Fast and Accurate Supervised Topic Models
Topic models provide insights into document collections, and their supervised extensions also capture associated document-level metadata such as sentiment. However, inferring such models from data is often slow and cannot scale to big data. We build upon the “anchor” method for learning topic models to capture the relationship between metadata and latent topics by extending the vector-space rep...
متن کاملHeadstart for speech segmentation: a neural signature for the anchor word effect.
Learning a new language is an incremental process that builds upon previously acquired information. To shed light on the mechanisms of this incremental process, we studied the on-line neurophysiological correlates of the so-called anchor word effect where newly learned words facilitate segmentation of novel words from continuous speech. Higher segmentation performance was observed for speech st...
متن کامل