Reducing the sampling complexity of topic models

By Aaron Q. Li, Amr Ahmed, Sujith Ravi, Alex Smola. 2014

224 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Natural Language Processing, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Philip S. Yu
2. Chuan Shi
3. Jian Tang
4. Jure Leskovec
5. Jun Zhu
6. Wayne Xin Zhao
7. Meng Qu
8. Charu C. Aggarwal

In Natural Language Processing

1. Jordan L. Boyd-Graber
2. Manzil Zaheer
3. Philip S. Yu
4. Rajarshi Das
5. Wayne Xin Zhao
6. Chris Dyer
7. Chuan Shi
8. Kaiwei Li

In Signal Processing

1. Chuan Shi
2. Jordan L. Boyd-Graber
3. Rajarshi Das
4. Philip S. Yu
5. Jun Zhu
6. Binbin Hu
7. Jian Tang
8. David Mimno

Citations in the corpus

• LINE: Large-scale Information Network Embedding (2015, methodology): “We use the alias table method [9] to draw a sample according to the weights of the edges, which takes only O(1) time when repeatedly drawing samples from the same discrete distribution.”
• WarpLDA: a Simple and Efficient O(1) Algorithm for Latent Dirichlet Allocation (2015, background): “With WarpLDA users can learn up to one million topics from hundreds of millions of documents in a few hours, at the speed of 2G tokens per second, or learn topics from small-scale datasets in seconds.”
• Gaussian LDA for Topic Models with Word Embeddings (2015, methodology): “In particular, we compute the document-specific sparse term exactly and for the remainder language model term we borrow idea from (Li et al., 2014).”
• Adversarial Network Embedding (2017, methodology): “To improve efficiency, the alias table method (Li et al. 2014) is employed to sample node from the candidate node set in every sampling step.”
• Applications of Topic Models (2017, background): “[2007], Liu et al. [2009], conditioning topics on authors and the communities the authors belong to.”
• WarpLDA: a Cache Efficient O(1) Algorithm for Latent Dirichlet Allocation (2015, methodology): “Let ᾱ = ∑K k=1 αk and β̄ = V β.”
• Learning Graph-based POI Embedding for Location-based Recommendation (2016, methodology): “In the latter procedure, we use alias table method [9] to draw a sample, thus reduce the sampling complexity to O (1).”
• Leveraging Meta-path based Context for Top- N Recommendation with A Neural Co-Attention Model (2018, methodology): “Given a node, we offline filter out all the out-going nodes with a low priority score, and further build an alias table [17] for O (1)-time node sampling.”