Statistical Analysis Stack Exchange
Statistical Analysis Stack Exchange

Questions tagged [structured-prediction]

13 questions
7
votes
1 answer

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 arbitrary graphs on surfaces, etc...). For…
user3658307
  • 1,754
  • 1
  • 13
  • 26
6
votes
1 answer

predicting tree structure

This topic is actually rather hard to google for as 'tree' has been overloaded in this domain to refer to decision trees. I'd be interested in having a learning algorithm produce code, such as used in Microsoft Power BI's feature to query databases…
Kiara Grouwstra
  • 161
  • 3
3
votes
1 answer

Any reasons to prefer neural networks over boosting methods in tabular data?

Based on Kaggle winners data, it seems that ensemble boosting methods like XGBOOST, LIGHTGBM, CATBOOST are the top choices when dealing with structured or tabular data for maximizing the prediction accuracy. However, in industry as far as I know,…
mhsnk
  • 118
  • 7
2
votes
0 answers

Any research on learning Bayesian network structure with a limit on the parent set size?

Learning a maximum-scored Bayesian network structure with bounded treewidth is rather popular in recent years, as stated in the paper A survey on Bayesian network structure learning from data in 2019. However, I failed to find out any recent…
Mengfan Ma
  • 203
  • 1
  • 5
2
votes
1 answer

Binary linear SVM in terms of structured SVM

On slide 32/85 of this tutorial on structured SVM learning, the author formulates binary SVM classification in terms of structured output SVM by defining $\Psi(x, y) = \frac{y}{2}x$. Why is there a constant factor of $1/2$ in the feature map and…
jogloran
  • 123
  • 5
1
vote
0 answers

How to model a problem where the aim is to predict if atleast one of 3 inputs is "different" from the rest?

I have a tabular dataset with 10000 rows, each with a RowID and 30 numerical features. There are multiple rows with the same RowID. The aim is to come up with a model, which at test time would take 3 rows (without the RowID) and predict if at least…
Train Heartnet
  • 123
  • 5
1
vote
0 answers

About Feature Engineering Tips When “discriminative algorithm care about no modeling the probability of the language”

I was going over my old NLP course slides and one of the pages is about using Structured Perceptron for tagging. It claims that because the algorithm is discriminative, it doesn't care about modeling the probability of the language, and thus "every…
Yan Yang
  • 111
  • 4
1
vote
0 answers

Regression - many samples have the same target

I have a machine learning problem in which I have a many-to-one relationship from samples to targets. I have ~3k samples but only 11 targets with a shared key YEAR My first approach was to reshape the targets to get a one-to-one relationship, but…
TobSta
  • 111
  • 1
1
vote
0 answers

How to encode dependence structure in regression (Bayesian networks)

Suppose you know the model should be of the form y =f(x1)g(x2,x3), where f and g are the functions I'm trying to find. Essentially, x2, x3 collectively predict some hidden variable z, and the response variable y is z dampened by some function of x1.…
Nezo
  • 131
  • 2
1
vote
0 answers

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, activations and optimiser. My current setup…
Tom Bolton
  • 11
  • 1
1
vote
0 answers

Which model for this information extraction problem?

I am trying to solve the following pattern recognition / information extraction problem. Assume I have a text where each token has been annotated by a single class among $K$ classes available (with a conditional random field for example). Results…
mic
  • 3,848
  • 3
  • 23
  • 38
1
vote
0 answers

Linear CRF is to LR what _____ is to RF/SVM

This figure from (1) shows the relation between LR and linear CRF: linear CRF is pretty much the generalization of logistic regression to sequences: Is there any established generalization of random forest or SVM to sequences? (1) Sutton, Charles,…
Franck Dernoncourt
  • 42,093
  • 30
  • 155
  • 271
0
votes
1 answer

How to take advantage of the information/structure we have in the labels in multiple output regression?

I have a regression problem where each observation possesses a vector of features and 4 associated responses. These responses, as in many problems are correlated. It would be nice to be able to predict them "all at once" in order to take into…
Tom
  • 1,204
  • 8
  • 17