Greedy learning of latent tree models for multidimensional clustering

Liu, Tengfei; Zhang, Nevin L.; Chen, Peixian; Liu, April Hua; Poon, Leonard K. M.; Wang, Yi

doi:10.1007/s10994-013-5393-0

Cited by 25 publications

(38 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…All clustering-based methods rely on two key points: grouping variables to identify new latent variables and constructing models in a hierarchical manner using a bottom-up strategy. Three types of structures have been studied: HLCMs via the bridged-islands (BI) algorithm [13], binary forests (in which each tree node can have at most two children) via the BIN-G and BIN-A algorithms [16], and non-binary forests (in which each tree node has no restrictions on the number of children) via the CFHLC algorithm [17]. These algorithms are all limited to work with discrete data.…”

Section: A Latent Tree Modelsmentioning

confidence: 99%

See 1 more Smart Citation

Incremental Learning of Latent Forests

2020

View full text Add to dashboard Cite

In the analysis of real-world data, it is useful to learn a latent variable model that represents the data generation process. In this setting, latent tree models are useful because they are able to capture complex relationships while being easily interpretable. In this paper, we propose two incremental algorithms for learning forests of latent trees. Unlike current methods, the proposed algorithms are based on the variational Bayesian framework, which allows them to introduce uncertainty into the learning process and work with mixed data. The first algorithm, incremental learner, determines the forest structure and the cardinality of its latent variables in an iterative search process. The second algorithm, constrained incremental learner, modifies the previous method by considering only a subset of the most prominent structures in each step of the search. Although restricting each iteration to a fixed number of candidate models limits the search space, we demonstrate that the second algorithm returns almost identical results for a small fraction of the computational cost. We compare our algorithms with existing methods by conducting a comparative study using both discrete and continuous real-world data. In addition, we demonstrate the effectiveness of the proposed algorithms by applying them to data from

show abstract

Section: A Latent Tree Modelsmentioning

confidence: 99%

“…Furthermore, they allow for exact probabilistic inference in linear time [3]. For this reason, LTMs have proven to be valuable in many areas, such as classification [4]- [6], topic detection [7], probabilistic inference [8] and cluster analysis [9]- [13].…”

Section: Introductionmentioning

confidence: 99%

Incremental Learning of Latent Forests

2020

View full text Add to dashboard Cite

show abstract

“…Instead, algorithms are evaluated empirically using held-out likelihood. It has been shown that, on real-world datasets, better models can be obtained using methods developed in this setting than using those developed in the first setting [26]. The reason is that, although the assumption of the first setting is reasonable for data from domains such as phylogeny, it is not reasonable for other types of data such as text data and survey data.…”

Section: Related Workmentioning

confidence: 99%

“…The state-of-the-art in this direction is an algorithm named EAST [25]. It has been shown [26] to find better models that alternative algorithms such as BIN [39] and CLRG [24]. However, it does not scale up.…”

Section: Learning Flat Modelsmentioning

confidence: 99%

Latent tree models for hierarchical topic detection

Chen

Zhang

Liu

et al. 2017

Artificial Intelligence

Self Cite

View full text Add to dashboard Cite

We present a novel method for hierarchical topic detection where topics are obtained by clustering documents in multiple ways. Specifically, we model document collections using a class of graphical models called hierarchical latent tree models (HLTMs). The variables at the bottom level of an HLTM are observed binary variables that represent the presence/absence of words in a document. The variables at other levels are binary latent variables, with those at the lowest latent level representing word co-occurrence patterns and those at higher levels representing co-occurrence of patterns at the level below. Each latent variable gives a soft partition of the documents, and document clusters in the partitions are interpreted as topics. Latent variables at high levels of the hierarchy capture long-range word co-occurrence patterns and hence give thematically more general topics, while those at low levels of the hierarchy capture short-range word co-occurrence patterns and give thematically more specific topics. Unlike LDA-based topic models, HLTMs do not refer to a document generation process and use word variables instead of token variables. They use a tree structure to model the relationships between topics and words, which is conducive to the discovery of meaningful topics and topic hierarchies.

show abstract

“…Extensive empirical studies have been conducted where the different algorithms are compared in terms of the BIC scores of the models they obtain and running time [8,9]. The experiments indicate that the EAST (Extension-Adjustment-Simplification-until-Termination) algorithm [10] finds the best models on data sets with dozens to around one hundred observed variables, while the BI (Bridged-Islands) algorithm [9] finds the best models on data sets with hundreds to around one thousand observed variables. On data sets with dozens to around one hundred variables, BI is much faster than EAST, while the models it obtains are sometimes inferior.…”

Section: Latent Tree Analysismentioning

confidence: 99%

A data-driven method for syndrome type identification and classification in traditional Chinese medicine

Zhang

Liu

et al. 2017

Journal of Integrative Medicine

View full text Add to dashboard Cite

The efficacy of traditional Chinese medicine (TCM) treatments for Western medicine (WM) diseases relies heavily on the proper classification of patients into TCM syndrome types. The authors developed a data-driven method for solving the classification problem, where syndrome types were identified and quantified based on statistical patterns detected in unlabeled symptom survey data. The new method is a generalization of latent class analysis (LCA), which has been widely applied in WM research to solve a similar problem, i.e., to identify subtypes of a patient population in the absence of a gold standard. A well-known weakness of LCA is that it makes an unrealistically strong independence assumption. The authors relaxed the assumption by first detecting symptom co-occurrence patterns from survey data and used those statistical patterns instead of the symptoms as features for LCA. This new method consists of six steps: data collection, symptom co-occurrence pattern discovery, statistical pattern interpretation, syndrome identification, syndrome type identification and syndrome type classification. A software package called Lantern has been developed to support the application of the method. The method was illustrated using a data set on vascular mild cognitive impairment.

show abstract

Greedy learning of latent tree models for multidimensional clustering

Cited by 25 publications

References 24 publications

Incremental Learning of Latent Forests

Incremental Learning of Latent Forests

Latent tree models for hierarchical topic detection

A data-driven method for syndrome type identification and classification in traditional Chinese medicine

Contact Info

Product

Resources

About