Questions tagged [conditional-random-field]
The conditional-random-field tag has no usage guidance.
62
questions
0
votes
1
answer
188
views
Computing the partition function using the forward-backward algorithm for linear chain CRFs
I'm trying to implement the forward-backward algorithm for a Linear Chain Conditional Random Field, as to compute the marginal distribution over labels for each time step in a sequence. I'm following ...
- 627
2
votes
0
answers
35
views
Conditional Gaussians in Infinite Dimensions
The law of the conditional Gaussian distribution (the mean and covariance) are frequently mentioned to extend to the separable Hilbert spaced valued case, i.e., for $(X,Y)$,
$$
\mu_{X|Y=y} = \mu_X - ...
- 159
3
votes
2
answers
136
views
Can I simulate a random field with nested variogram by a summation of independently simulated random field for each componnet of the nested variogram?
I want to simulate a random field $Z(u)$ that has a nested variogram, say $\gamma(h)=\gamma_1(h) + \gamma_2(h) + \gamma_3(h)$, assuming the variogram is isotropic. Whether can I simuate independently ...
- 511
2
votes
1
answer
484
views
Conditional posterior distribution of coefficient in logistic regression
I would like to derive the conditional distribution of β in a logistic regression where
Y follows a Binomial distribution B(n,p)
the probability model is given as: logit(p)=X'β +u+v ;
u is spatial ...
- 21
0
votes
0
answers
341
views
Computing the gradient of the log-partition function in a linear-chain conditional random field (CRF) model
Query.
When computing the gradient of the log-partition function for an exponential family distribution specified by the linear-chain conditional random field (CRF) model, will unary conditional ...
- 2,425
1
vote
1
answer
367
views
Can I transform the Viterbi score from tensorflow into a probability?
I understand the Viterbi algorithm as it is explained in Wikipedia
However, the TensorFlow implementation is different:
...
- 13
1
vote
0
answers
115
views
Are CRF/MRF/GRF still used widely in computer vision?
I've tried to find recent (the year 2020) popular works that use Markov/Gibbs/Conditional Random Fields. My approach was: go to Google Scholar and find the works, citing a few relevant works on this ...
- 121
2
votes
1
answer
93
views
Operator notation - numerically stable forward recursions in linear-chain conditional random fields (CRFs)
I would appreciate some clarity on my understanding of some notation for the mathematical specification of forward recursions for inference in linear-chain conditional random fields (CRFs), with a ...
- 2,425
0
votes
1
answer
425
views
difference between Conditional Random Fields(CRF) and Naive Bayes
I am working through the course Probabilistic Graphical Models.
A CRF calculates the conditional probability distribution, i.e P(Y | X1, .., XN). The following picture shows an example how the CRF is ...
- 101
1
vote
1
answer
288
views
unary potentials in graph cuts like optimization for computer vision
I am currently reading a paper which performs minimisation of energy function of the form:
$
E(z, \alpha) = U(z, \alpha) + \lambda V(z, \alpha)
$
Here $z$ is the image data defined in a uniform grid ...
- 4,580
2
votes
2
answers
2k
views
What are the advantages of combining BiLSTM and CRF?
BiLSTM-CRF is a common model for sequence tagging (POS tagging, NER, ect.). What are the advantages of combining BiLSTM and CRF? What is the role of each one of the parts in this combination?
- 155
2
votes
2
answers
595
views
Named entity recognition with only one pure entity(no context)?
We know that we can extract entities from a sentence using named entity recognition, but what if the sentence contains only an entity and no other context?
For example, we can use CRF for the ...
- 6,148
11
votes
1
answer
2k
views
Relation between Gaussian Processes and Gaussian Markov Random Fields
As a non expert in the field, I am relating Gaussian Processes (GP) and Gaussian Markov Random Fields (GMRF).
I might just be confused by the fact that different resources use different formalism. ...
- 353
1
vote
1
answer
110
views
The implementation of variable-to-factor and factor-to-variable messages?
I read this tutorial on the implementation of CRF and got to know that the normalization is the sum-product message passing. And I also know that there are two types of messages on factor graph:
...
- 6,148
2
votes
1
answer
358
views
The difference between forward algorithm used in CRF and the variable elimination?
I found that in the forward algorithm used in the CRF(and perhaps also in the HMM) the mechanism applied is almost the same as that in the variable elimination(VE) except that the emission ...
- 6,148
1
vote
1
answer
838
views
Introduction to Conditional random fields
I came across the application of a conditional random field (CRF) to the output from a convolutional neural network (CNN) for image segmentation. The additional CRF step seems to be a common ...
- 559
2
votes
0
answers
1k
views
loss function in CRF keras-contrib returns Nan in join mode
I use a BiLSTM-CRF architecture to assign some labels to a sequence of the sentences in a paper. We have 150 papers each of which contains 380 sentences and each sentence is represented by a double ...
- 21
3
votes
0
answers
48
views
Sequential Prediction: Data Modeling and Classical Algorithms
I have data that can be called demographic data.
Raw data
Person 0001
\begin{array}{|c|c|}
\hline
Feb\,1981- Apr\,85 & engaged\,\,in\,\,\underline{activity}\,\,\textit{A}\,\,of \,\,\underline{...
0
votes
1
answer
175
views
Does Alphabet Case Matters while Training Stanford Open NLP NER classifier?
I'm working with Named Entity Recognition for non-English. I've some raw text file (all small letter) and trying to make NER classifier. I'm not sure If it'll be better using Small Capital mixed text ...
- 11
1
vote
0
answers
183
views
LSTM with only one node in the hidden layers
I have a regression problem where I'd like to train an LSTM to tackle it. However, since I do not have too many samples for training (only 2000), I am thinking about using only one hidden node, since ...
- 121
1
vote
1
answer
1k
views
NER at sentence level or document level?
Should NER models (LSTM or CRF) take input training data at sentence level or paragraph level?
Let's say we have this input text, and we would like to do Named Entity Extraction:
GOP Sen. Rand ...
- 63
2
votes
0
answers
31
views
Noisy conditional simulation
A conditional random field $Z_C(x)$ is a random field whose realisations $z_C(x)$ always take the same values $z_C(x_a)$ at locations $x_a$. Realisations of $Z_C(x)$ can be produced as follows (...
- 1,165
4
votes
1
answer
731
views
HMM and CRF: the label bias problem and I-equivalence
I have a question about the label bias problem in HMM and CRF.
I understand that HMM and MEMM suffer from the label bias problem, which is a preference for states with fewer transitions. The problem ...
- 668
1
vote
0
answers
27
views
Options for spatially correlated outputs (i.e. structured regression)?
I have a convolutional neural network which takes a 40x40 real valued input, and maps it to a real valued 40x40 output. I've optimized the number of convolution layers, filter size, hidden layer size, ...
- 11
1
vote
0
answers
2k
views
CRF (TensorFlow) log-likelihood becomes positive
I am currently doing a multiclass classification task on sequence data and am using tf.contrib.crf.crf_log_likelihood to compute sentence level log-likelihood values.
In particular it implements a ...
- 93
6
votes
2
answers
622
views
Python package that allows to train a CRF on two datasets
I am looking for a Python package that allows to train a conditional random field (CRF) on two datasets.
For example: I have two datasets, dataset A and dataset B. I want to train a conditional ...
- 45.5k
3
votes
1
answer
2k
views
derivation of partition function in conditional random fields
When reading the paper of Efficient piecewise training of deep structured models for semantic segmentation, I am confused about the derivation in CRF training (section 6). In specific, I do not know ...
- 1,307
1
vote
0
answers
47
views
Including document level information in sequence tagging
Typically sequence taggers (specifically, linear-chain CRFs) in NLP use sentence level information, i.e., for each word, we define feature functions that only depend on the tokens in the sentence. I ...
- 424
2
votes
1
answer
85
views
Why should the last tag be fixed in CRF?
In the tutorial, it is said that:
Let v range over the tags. Define U(k, v) to be the score of the best
sequence of tags from 1 to k, where tag k is required to be v. This is
a maximization ...
- 6,148
0
votes
0
answers
280
views
Taking forever to converge
I'm using the HCRF Library for a sequence prediction task . For learning the model parameters, I'm using LBFGS, although there are options to use CG and BFGS.
I notice that it take an awfully long ...
- 729
5
votes
2
answers
2k
views
How can I implement a CRF feature function?
It is said that a feature function can represent anything, the first or the last word of a sentence, a capital character and etc. But how exactly can I represent them in such a form: $F_j(x, y)$ or $\...
- 6,148
2
votes
0
answers
153
views
CRF message passing as convolution operation
I was reading this particular paper: Efficient Inference in Fully Connected CRFs with Gaussian Edge Potentials, and I didn't understand this equation (eq 5) in the paper:
I understand the first ...
- 143
2
votes
0
answers
56
views
What is the connection between CRF and Logistics Regression? [duplicate]
Logistic regression seems to be a simple version of Conditional Random field, I am unable to figure out how. Perhaps my fundamentals are shaky.
- 559
1
vote
1
answer
253
views
Fast Gaussian Filtering Using a Permutohedral Lattice
I'm trying to implement
http://www.dabi.temple.edu/~zoran/papers/KostaAAAI13.pdf
but am stack at understanding equations 10) and 11).
They claim that the sum of the Gauss kernel multiplied with the ...
- 191
7
votes
0
answers
604
views
Deep Learning vs Structured Learning
I am interested in the differences between using large, deep learning networks vs Probabilistic graphical models (PGMs), like Random Field models, for structured learning (e.g. on images, or labels of ...
- 2,174
1
vote
1
answer
2k
views
CRF equivalent in deep learning
Conditional Random Fields (CRFs) is a typical solution for a sequence labelling/segmentation problem. For example, a sequence is a string and CRFs are used to label each word as being a part of a ...
- 31
1
vote
0
answers
28
views
Using Conditional random field for many valued labels
I want to use the CRF for labeling a corpus of annotated text. Each word in the corpus has its own set of labels. More specifically, the labels are the pronunciations of each word: some words like "...
- 163
2
votes
0
answers
531
views
unsupervised methods for conditional random fields (CRF)
can any one explain that why CRFs are not applicable for unsupervised learning?
thanks in advance
- 21
2
votes
1
answer
749
views
A Question on forward-backward algorithm in CRF++
I’m a beginner of CRF++. A question have plagued me for many days. I’m exhausted~
Why the code in calcBeta() is so like ...
- 23
2
votes
1
answer
368
views
Is there any advantage of using MEMM instead of CRF for named-entity recognition?
I wonder whether there is any advantage of using maximum-entropy Markov model (MEMM), a.k.a. conditional Markov model (CMM) instead of using conditional random fields (CRF) for named-entity ...
- 45.5k
1
vote
0
answers
497
views
Stacked Conditional Random Field (CRF) implementation
I am looking for a library that can train a stacked conditional random field (CRF). I plan to use it for natural language processing purposes. Ideally, Python interface, works on Linux, and multi-...
- 45.5k
2
votes
1
answer
737
views
CRF implementation with Python API that allows a CRF model to be trained multithreadedly
I am looking for a conditional random field (CRF) implementation with a Python API that allows a CRF model to be trained multithreadedly.
I currently use pyCRFsuite, which works great except that CRF ...
- 45.5k
5
votes
1
answer
1k
views
Sequence length when training a conditional random field (CRF)
I am training a conditional random field (CRF) to perform named entity recognition. I have 1000 documents, each containing from 100 to 500 sentences.
During the training phase, is it better to train ...
- 45.5k
3
votes
0
answers
168
views
What means random field in "Conditional Random Field" [closed]
I have studied about CRF but the popular papers and tutorials have not describe the philosophy of CRF. I see the terms "Field", "Random Field" and "Conditional" from terminological aspect. But the ...
- 173
3
votes
0
answers
35
views
Sequence tagging with additional structure
I am looking for pointers (papers, algorithms etc) for learning models for sequence tagging but which allow for additional structure. Consider Part of Speech Tagging, I could train a CRF which would ...
- 2,424
1
vote
0
answers
87
views
Compute partition function
I'm given a distribution on 3 discrete variables $x, y, z$ which is defined as
$$p(x, y, z) = \frac{1}{Z} \psi(x, y, z) = \frac{1}{Z} \phi_1(x, y) \phi_2(y, z)$$
where $x, y$ can take up value among ...
- 11
2
votes
1
answer
57
views
How can I solve this graphical model?
I have a classification problem, with the following structure. There is a fully-connected graph, and each node needs be assigned a class label. Every pair of nodes in the graph has a probability ...
- 6,749
2
votes
1
answer
1k
views
What is the number of features in CRFsuite / python-crfsuite? [closed]
I wonder what the number of features is in python-crfsuite
.
I thought that the number of features was the number of attributes multiplied by the number of labels, e.g.:
...
- 45.5k
3
votes
1
answer
891
views
Generative-Discriminative pairs: Naive Bayes and Logistic Regression
I'm trying to understand the something written in this paper.
At the bottom of page 7:
This means that if the naive Bayes model $$ p(y,\mathbf{x}) = p(y) \prod_k p(x_k|y) $$ is trained to
...
- 737
4
votes
0
answers
483
views
Is every Conditional Random Field simply a Markov Random Field with restricted structure?
If I have a graph $H$ with nodes $\mathbf{X} \cup \mathbf{Y}$, and a set of factors $\phi_1(D_1), \ldots, \phi_k(D_k)$, where for each $i$, $D_i \not\subset X$, then doesn't this define both a MRF and ...
- 737