Find research datasets worth reusing

Figure 6. Two point crossover of HMM chromosomes.-Neuroevolution Mechanism for Hidden Markov Model

Two point Crossover For the two point crossover we get two parent HMMs and choose at random two cutting points for the weights that have a sum of 1.0 and swap the contents between the crossing points. This is illustrated in the example shown in Figure 6.

Figure 4. A chromosome structure for HMM shown in Figure 2.-Neuroevolution Mechanism for Hidden Markov Model

The chromosome which represents the HMM can be extracted from its corresponding neural network. The general structure of the chromosome is divided into two sections, input layer and hidden layer. Each section contains many slots, and each slot represents a weight from one node in that layer to a node in the next layer (from input to hidden and from hidden to output). The number of slots in the input layer is the same number of input nodes in the neural network. In the hidden layer, number of slots is equal to nodes in the output layer multiplied by the nodes in the hidden layer.

Figure 3. Neural network representation for HMM given in Figure 2.-Neuroevolution Mechanism for Hidden Markov Model

In our proposed structure, we injected a hidden layer to have a multilayer perceptron which is more efficient than single layer perceptron. To make this process clear, Figure 3 shows the neural networks for the HMM presented in Figure 2.

Figure 1. HMM to describe a relation between the states Med. and High with the observations (invisible states) cold and hot.-Neuroevolution Mechanism for Hidden Markov Model

The advantage of using this technique is that MCPRs are very useful in real time applications and can be adapted over time based on the obtained experience of the networking working process. Again Hewahi[6] proposed a mechanism (algorithm) to evolve and select the best suitable HMM for a given problem using GA, this mechanism lacks to the training process that can be of great usefulness in finding the best HMM. Based on the above mentioned research, the importance of using HMM is increasing rapidly. Let us consider the HMM presented in Figure 1.

Figure 2. HMM with weights and necessary conditions on top of edges-Neuroevolution Mechanism for Hidden Markov Model

Based on the HMM structure in Figure 2, we can perform the following steps: 1. Make the number of nodes of inputs in the input layer of the NN as the number of states (visible states not the observations). Each input node represents one state. 2. Number of nodes in the output layer in the NN is equal to the number of states and observations (visible and invisible states), where each node corresponds to one state (visible or invisible). 3. We construct a hidden layer in NN with n number of nodes, where n is the same number of nodes in the input layer. 4. We make a connection from every input to every hidden layer node with a very negligible weight. 5. Connect every hidden node in the hidden layer to every node in the output layer. 6. Assign weights from the hidden layer to output layer in a way that as every node in the hidden layer corresponding to input state. The weight on top of the link between the hidden node to the output node is the probability value between the states in the HMM. In our proposed structure, we injected a hidden layer to have a multilayer perceptron which is more efficient than single layer perceptron. To make this process clear, Figure 3 shows the neural networks for the HMM presented in Figure 2.

Figure 7. One point crossover of HMM chromosomes.-Neuroevolution Mechanism for Hidden Markov Model

This crossover is performed in the input layer part only. We choose a crossing cut point in the input layer part of the chromosome, and exchange everything before it. This is illustrated in Figure 7.

Figure 3. The crossover operation between two HMM chromosomes Figure-Genetic Algorithms Principles Towards Hidden Markov Model

Crossover In this genetic operator, we choose two chromosomes at random and apply crossover between them. Figure 3 shows the proposed crossover. We choose a crossing cut site at random. It is to be noted that the crossing cut site should be even number. We should have two crossing cut sites. If we make crossing cut site odd number, the resultant child will not have a correct value of probability. The incorrect crossover is shown in Figure 4.

Figure 1. HMM to describe a relation between the states Med. and High with the observations (invisible states) cold and hot.-Genetic Algorithms Principles Towards Hidden Markov Model

Hewahi [4] presented a modified version of Censored Production Rule (CPR) called Modified Censored Production Rules (MCPR). CPR is proposed by Michalski and Winston [6 ] to capture real time situations. MCPR can fit with hidden Markov model and present a scheme to compute the certainty values of the obtained conclusions out of the induced rules. To compute the certainty values for the rule actions (conclusions), the approach exploited only the probability values associated with the hidden Markov model without using any of the other well known certainty computation approaches. Hewahi [3] also proposed an intelligent networking management system based on the induced MCPRs extracted from a networking structure based on HMM. The advantage of using this technique is that MCPRs are very useful in real time applications and can be adapted over time based on the obtained experience of the networking working process. Let us consider the HMM presented in Figure 1.

opencc-by-4.0Sep 2023View details →

Sequence and HMM libraries

This repository contains the data specified in the paper entitled "A refined picture of the native Amine Dehydrogenase family revealed by extensive biodiversity screening". It includes: 1. NAD-dependent_enzymes.fa.gz - The library of 20,315,745 sequences of NADPH-dependent enzymes recovered from genomic and metagenomic sequence databases. 2. ref-AmDHs17959_nr.fa.gz - The library of 17,959 ref-AmDH sequences recovered from genomic and metagenomic sequence databases. Considered as the updated nat-AmDH family. 3. NAD_subfams_HMMs.tar.gz - The library of 104,686 Hidden Markov Models (HMMs) of NADPH-dependent protein subfamilies. As described in the paper, those HMMs were obtained by clustering the set uploaded here as NAD-dependent_enzymes.fa.gz and building one HMM per subfamily. 4. ref-AmDHs_HMMs.tar.gz - This repertory includes the HMMs used to update the nat-AmDH family (all_ASMC_no_nad,hmm and all_ASMC_nad_dom.hmm) as well as the ones used to search for distant homologs; HMMs designed for the phylogenetic and structure-based groups built from the set uploaded here as ref-AmDHs17959_nr.fa.gz (asmc_*.hmm and phylo_*.hmm) . 5. ref-AmDH_ASMC_models.tar.gz - The library of 9886 ref-AmDH models built using the ASMC pipeline. 6. 72_ref-AmDH_seqs_representatives.txt.gz - Sequences of the 72 representative ref-AmDHs experimentally tested and found to be active. 7. 17_nat-AmDH_seqs_specific_feature.txt.gz - Sequences of the 17 nat-AmDHs with specific feature that have been heterologously expressed and tested.

zenodo36/100

Composite Embedding Systems Based on DNN-HMM and Attention End-To-End for ZeroSpeech2017 track1 (1)

Deep neural networks (DNNs) were trained for posterior and bottleneck features using Japanese and other language speech data. We explore various DNN types, their combinations, and dimension reduction by principal component analysis (PCA). This version (version 1) extracts DNN bottleneck features obtained from GMM based SAT features. The DNN and GMM were trained by speech data from the corpus of spontaneous Japanese (CSJ).

opencc-by-4.0Jun 2017View details →

zenodo36/100

Composite Embedding Systems Based on DNN-HMM and Attention End-To-End for ZeroSpeech2017 track1 (2)

Deep neural networks (DNNs) were trained for posterior and bottleneck features using Japanese and other language speech data. We explore various DNN types, their combinations, and dimension reduction by principal component analysis (PCA). This version (version 2 ) concatenates CSJ feature vector and PCA compressed feature vector made from attention end-to-end feature. X:CSJ feature (60 dim bottleneck, (version 1 feature)) S:Attention end-to-end feature (320 dim) T:PCA(S) (60 dim) Z=concat(X,T)

opencc-by-4.0Aug 2020View details →

RVDB-prot, a reference viral protein database and its HMM profiles

We present RVDB-prot, a database corresponding to the protein equivalent of the nucleic acid reference virus database RVDB. Protein databases can be helpful to perform more sensitive protein sequence comparisons. Similarly to its homologous public repository, RVDB-prot aims to provide reliable and accurately annotated unique entries, while including also an Hidden Markov Model (HMM) protein profiles database for distant protein searching.

Deep Learning Methods for Unsupervised Acoustic Modeling using HMM posteriograms (system #2)

System combination of HMM-DNN with auto encoder features

Unsupervised Acoustic Modeling using Autoencoder-DNN with HMM Posteriograms (system #3)

DNN trained with Autoencoder features with HMM posteriograms.

Deep Learning Methods for Unsupervised Acoustic Modeling using HMM posteriograms

DNN trained using HMM posteriograms

opencc-by-4.0Mar 2022View details →

Metascan metabolic HMM database and auxilliary files

Metascan HMM database and auxilliary files, designed by clustering (large sets of) metabolic proteins with similar functions before creating HMM profiles.