Variational Bayesian Learning of Probabilistic Discriminative Models With Latent Softmax Variables

Ahmed, Nisar; Campbell, Mark

doi:10.1109/tsp.2011.2144587

Cited by 19 publications

(5 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…After the vector representation with lexical weight is calculated through using the attention mechanism, it is spliced and input into a dimension reduction full connection layer and then classified by using the SoftMax function [12].…”

Section: A Topic Segmentation Model Based On Model Transfermentioning

confidence: 99%

Research on the Semantic Analysis Method of Translation Corpus Based on Natural Language Processing

Yue

2022

Scientific Programming

View full text Add to dashboard Cite

Accurate recognition and analysis of semantics is the most important research field in the process of English translation with the help of natural language processing technology. This paper proposes an English semantic analysis method based on the neural network. First, the idea of model transfer is used to construct a topic segmentation model and the topic granularity segmentation of the translated text is carried out. Then, in order to obtain all the information in the English text, the recursive neural network is selected to recognize the word model. In order to recognize English texts with different sentence patterns, the long-term and short-term memory network is selected to extract the useful information of the text. Through the experimental data measurement and analysis results, compared with the traditional sentence analysis methods, the accuracy of the proposed method is as high as 95.8% and the model occupies less hardware resources.

show abstract

Section: A Topic Segmentation Model Based On Model Transfermentioning

confidence: 99%

Research on the Semantic Analysis Method of Translation Corpus Based on Natural Language Processing

Yue

2022

Scientific Programming

View full text Add to dashboard Cite

show abstract

“…It is assumed that the detection rate is high enough and ζ k is highly accurate so that observation uncertainty and data association ambiguity for this type of observation do not need to be considered. For the likelihood function p(ζ k |x), a 2D multi-modal softmax (MMS) model [35] shown in Fig. 7 is used, and weights and biases of the MMS model are calculated offline [26] and are modified online based on the rover's position and orientation.…”

Section: A Problem Setupmentioning

confidence: 99%

Probabilistic Semantic Data Association for Collaborative Human-Robot Sensing

Wakayama¹,

Ahmed

2021

Preprint

Self Cite

View full text Add to dashboard Cite

In collaborative human-robot semantic sensing problems, e.g. for scientific exploration, robots could potentially overtrust information given by a human partner, resulting in suboptimal state estimation and poor team performance. When humans cannot be treated as oracles, robots need to update state beliefs to correctly account for possible discrepancies between human semantic observations and the actual world states which lead to those observations. This work develops strategies for rigorous online calculation of probabilistic semantic data association (PSDA) probabilities for semantic likelihoods in general settings, unlike previous work which developed naive or heuristic approximations for specific settings. The new PSDA method is incorporated into a hybrid Bayesian data fusion scheme which uses Gaussian mixture priors for object states and softmax functions for semantic human sensor observation likelihoods, and is demonstrated in Monte Carlo simulations of collaborative multi-object search missions featuring a range of relevant human sensing characteristics (e.g. false detection rate). It is shown that PSDA leads to robust estimation of observation association probabilities under a wide range of conditions whenever semantic human sensor data contain significant target reference ambiguities for autonomous object search and localization.

show abstract

“…The conditional distribution for the unknown noise variance R k − 1 at time k − 1 could be approximated as Inverse-Gamma distribution, because the Inverse-Gamma distribution is the conjugate prior distribution for the variance of a Gaussian distribution [17]. The Inverse-Gamma distribution for the noise variance could be represented as…”

Section: Analytic Implementation Of the Extended Phd Filtermentioning

confidence: 99%

“…To solve this problem, the true posterior distribution v k|k (x k , R k |Z 1:k ) can be approximated by a distribution that is computationally tractable, which called VB approximation [17][18][19]. The key idea of VB is to find a tractable approximation to the true posterior density that minimises the Kullback-Leibler (KL) divergence.…”

Section: Propositionmentioning

confidence: 99%

“…Under linear Gaussian assumptions on the target models, the posterior intensity for targets at time k − 1 is a Gaussian mixture of the form [16]

v_{k - 1} false(bold-italicx false) = \sum_{i = 1}^{J_{k - 1}} {bold-italicw}_{k - 1, i} N (x; {bold-italicm}_{k - 1, i}, {bold-italicP}_{k - 1, i})

The conditional distribution for the unknown noise variance R k − 1 at time k − 1 could be approximated as Inverse‐Gamma distribution, becau s e the Inverse‐Gamma distribution is the conjugate prior distribution for the variance of a Gaussian distribution [17]. The Inverse‐Gamma distribution for the noise variance could be represented as

p ()(, R_{k - 1} falsefalse| Z_{1 : k - 1}) = \prod_{j = 1}^{d} Inv ‐ Gamma ()(, {(σ_{k - 1}^{j})}^{2}; α_{k - 1}^{j}, β_{k - 1}^{j})

The superscript ‘ d ’ denotes the dimension of R k .…”

Section: Analytic Implementation Of Gaussian Mixture Phd Filter Usimentioning

confidence: 99%

See 1 more Smart Citation

Adaptive noise variance identification for probability hypothesis density‐based multi‐target filter by variational Bayesian approximations

Huang

Gao

2013

IET Radar, Sonar & Navigation

View full text Add to dashboard Cite

A new extended probability hypothesis density (PHD) filter is proposed for joint estimation of the time-varying number of targets and their states without the measurement noise variance. The extended PHD filter can adaptively learn the unknown noise parameters at each scan time by using the received measurements. With the decomposition of the posterior intensity separated into Gaussian and Inverse-Gamma components, the closed-form solutions to the extended PHD filter are derived by using the variational Bayesian approximations, which have been proved as a simple, analytically tractable method to approximate the posterior intensity of multi-target states and time-varying noise variances. Simulation results show that the proposed filter can accommodate the unknown measurement variances effectively, and improve the estimation accuracy of both the number of targets and their states.

show abstract

Variational Bayesian Learning of Probabilistic Discriminative Models With Latent Softmax Variables

Cited by 19 publications

References 30 publications

Research on the Semantic Analysis Method of Translation Corpus Based on Natural Language Processing

Research on the Semantic Analysis Method of Translation Corpus Based on Natural Language Processing

Probabilistic Semantic Data Association for Collaborative Human-Robot Sensing

Adaptive noise variance identification for probability hypothesis density‐based multi‐target filter by variational Bayesian approximations

Contact Info

Product

Resources

About