Incremental probabilistic Latent Semantic Analysis for video retrieval

Abstract

Recent research trends in content-based video retrieval have shown topic models as an effective tool to deal with the semantic gap challenge. In this scenario, this work has a dual target: (1) it is aimed at studying how the use of different topic models (pLSA, LDA and FSTM) affects video retrieval performance; (2) a novel incremental topic model (IpLSA) is presented in order to cope with incremental scenarios in an effective and efficient way.A comprehensive comparison among these four topic models using two different retrieval systems and two reference benchmarking video databases is provided. Experiments revealed that pLSA is the best model in sparse conditions, LDA tend to outperform the rest of the models in a dense space and IpLSA is able to work properly in both cases.

Hyperspectral Unmixing Based on Dual-Depth Sparse Probabilistic Latent Semantic Analysis

“…LDA potentially overcomes these drawbacks by using two Dirichlet distributions, one to model documents θ ∼ Dir(α) and another to model topics p(w|t,β) ∼ Dir(β). iterating over the document collection which results in LDA requiring relatively dense distributions to obtain a good hyper-parameter estimation [30]. Even authors in [31] Second, the proposed approach does not consider any kind of prior distribution but only the own data, which eventually simplifies the complexity of the model when compared to other Bayesian approaches that assume prior distributions with some hyperparameters [41], [42].…”

Multimodal Probabilistic Latent Semantic Analysis for Sentinel-1 and Sentinel-2 Image Fusion

“…However, this generative process is usually called ill-defined because documents set topic mixtures and simultaneously topics generate documents, thus there is not a natural way to infer previously unseen documents [8]. Additionally, the number of pLSA parameters grows linearly with the number of training documents which makes this model particularly memory demanding and susceptible to over-fitting [13].…”

Sparse multi-modal probabilistic latent semantic analysis for single-image super-resolution

“…According to the standard pLSA model complexity [38], sMpLSA-tst cost is , that is, the computational burden of the regular pLSA model. However, it is important to highlight that the sMpLSAtst model only has a single parameter to be estimated, i.e.…”