Topic regression multi-modal Latent Dirichlet Allocation for image annotation

By Duangmanee Putthividhya, Hagai Thomas Attias, Srikantan S. Nagarajan. 2010

206 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Multimodal Learning, Probabilistic & Bayesian Methods

How researchers see this paper

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

In Computer Vision and Pattern Recognition

In Multimodal Learning

In Probabilistic & Bayesian Methods

Citations in the corpus

Self-Supervised Learning of Visual Features through Embedding Images into Text Topic Spaces (2017, neutral): “Supervised variations of LDA are presented in [30, 42, 28] where the discovered topics are driven by the semantic regularities of interest for the classification task.”
A Comprehensive Survey on Cross-modal Retrieval (2016, methodology): “Corr-LDA [21], Tr-mm LDA [22], MDRF [23] Deep learning methods Multimodal Deep Autoencoder [24], Multimodal DBM [25], DCCA [26], End-to-end DCCA [27], Corr-AE [28], Joint Video-Language Model [29] Pairwise based methods”
Latent Dirichlet Allocation Models for Image Classification (2013, methodology): “Several other adaptations of LDA have been proposed for tasks other than classification: correspondence LDA [2] and topic-regression multimodal LDA [17] for image annotation, labeled LDA [19] (labLDA) for credit attribution in a multilabeled corpora, semiLDA [26] for human action recognition in vide”
Learning Tags from Unsegmented Videos of Multiple Human Actions (2011, background): “Topic models [26], [16], [22] based on CorrLDA [4] have been shown to be suited for this task, typically by learning a multinomial distribution over tags conditional on each topic.”
Joint Feature Selection and Subspace Learning for Cross-Modal Retrieval (2016, methodology): “To capture the correlation between images and annotations, Latent Dirichlet Allocation (LDA) [12] has been extended to learn the joint distribution of multi-modal data such as Correspondence LDA (Corr-LDA) [10] and Topic-regression Multi-modal LDA (Tr-mm LDA) [11].”
MMALFM (2018, methodology): “topics. In Corr-LDA [7], image is the primary modality and is generated first, and the caption word is then generated based on the topic of an image region which the word is associated with. Tr-mmLDA [50] uses a latent variable regression approach to learn a linear mapping between the topic distribu”
A Thousand Frames in Just a Few Words: Lingual Description of Videos through Latent Topics and Sparse Object Stitching (2013, methodology): “The param- eters for corpus level topic multinomials over textual words are β1:K—only the training instances of these parameters are used for keyword prediction.”
A Sentence Is Worth a Thousand Pixels (2013, methodology): “In [20], regression is used to predict the topics in one modality from the topics in the other.”
Supervised Coupled Dictionary Learning with Group Structures for Multi-modal Retrieval (2013, methodology): “…multi-modal data (e.g., texts and images) such as Correspondence LDA (Corr-LDA) (Blei and Jordan 2003), Topic-regression Multi-modal LDA (tr-mmLDA) (Putthividhy, Attias, and Nagarajan 2010), Multi-field Correlated Topic Modeling (mf-CTM) (Salomatin, Yang, and Lad 2009) and Hierarchical Dirichlet…”
Attentive Recurrent Neural Network for Weak-supervised Multi-label Image Classification (2018, background): “A novel topic-model based latent variable regression approach is presented [27] to capture correlations between image features and labels.”
Attribute Learning for Understanding Unstructured Social Activity (2012, methodology): “We have made no explicit model [27] of the different modalities of observations in our data.”
A Deep and Autoregressive Approach for Topic Modeling of Multimodal Data (2014, methodology): “As previously mentioned, multimodal data is often modeled using extensions of the basic LDA topic model, such as Corr-LDA [2], Multimodal LDA [5] and MDRF [4].”
Topic Modeling of Multimodal Data: An Autoregressive Approach (2014, methodology): “To deal with multimodal data, some variants of LDA have been proposed recently [2, 5, 4, 9].”
An Overview of Cross-Media Retrieval: Concepts, Methodologies, Benchmarks, and Challenges (2017, methodology): “topic model [97](U) [98](U) [99](U) [100](F)”
Learning cross-modality similarity for multinomial data (2011, methodology): “To generalize over this requirement for one-to-one topic correspondence, the topic regression multi-modal LDA model was recently proposed [13].”