Methods and principles of building "computer systems that automatically improve with experience."

learn more… | top users | synonyms (1)

1
vote
1answer
47 views

GridSearchCV and KFold

I noticed that in some cases, a GridSearchCV is applied on the output of KFold. For example, like in the code below. Why is it needed? I thought that something equivalent to KFold is already applied ...
1
vote
0answers
27 views

Tuning Parameters for Boosting/Bagging/Random Forest

I want to use tree-based classifiers for my classifiaction problem. I'm thinking about bagging, boosting (AdaBoost, LogitBoost, RUSBoost) and Random Forest but I'm unsure about the tuning parameters, ...
0
votes
0answers
8 views

Using priors, weights or costs for mitigating class imbalance?

A plethora of Matlab classifiers (e.g. tree-based or svm) allow to set priors, costs or weights for the data points. This can help dealing with imbalanced data. Unfortunately, none does support ...
2
votes
0answers
19 views

How to model timeseries temperature data?

I would like to model a timeseries consisting of internal temperature data of a greenhouse, collected at 15 min interval and then use it to make predictions in the future. This is how my data looks ...
0
votes
0answers
32 views

Keras - Convolution neural network accuracy drop

I am running a convolutional neural network in keras with the cfar-10 dataset. I have 3 layers: conv -> dense -> output. What is strange to me is that when it runs It starts giving me about 90% ...
1
vote
2answers
34 views

Similarity of two neural networks

I have two neural networks. If I take only weights (the activation functions for both are the same), is there a way to tell the percent similarity of these two networks?
0
votes
0answers
10 views

Do you need to scale Vectorizers in sklearn?

I have a set of custom features and a set of features created with Vectorizers, in this case TfidfVectorizer. All of my custom features are simple np.arrays (e.g. [0, 5, 4, 22, 1]). I am using ...
3
votes
0answers
27 views

classifier performing well in leave-one-out cross-validation but not k-fold

I am building a classifier and have over 1 million features to choose from. I implemented penalized regression, aka, lasso regression, followed by recursive feature selection in order to select ...
1
vote
1answer
87 views

Newton's method for regression analysis without second derivative

In regression analysis, instead of gradient descent, Newton's method can be used for minimizing the cost function. However, in Newton's method, we need to calculate second derivative too. For ...
0
votes
0answers
21 views

why am i getting bad predictions rates?

I am trying to make a classifier capable of recognizing digits using the naive bayes method. Problem is though that i am getting pretty bad results. I thought the reason would be because of the ...
0
votes
0answers
17 views

online learning- winnow algorithm and mistake bound

I came across an interesting question and I must say I am struggling to figure out how it suppose to work. So we consider the winnow algorithm that learns non-monotone disjunctions. Could someone ...
5
votes
3answers
114 views

Overview of predictive modelling, machine learning, etc.

My text Intuitive Biostatistics is a nonmathematical explanation of conventional statistics. Chapter 3 explains the basic mindset of statistics as analyzing a sample to make inferences from a ...
0
votes
0answers
18 views

Implementing an Adaboost Classifier

I have generated an adaboost classifier in Weka on a dataset where each instance falls into one of two classes. The result was a number of decision trees, each assigned a weight. What is the ...
0
votes
1answer
13 views

Classification based on a large number of independent continuous variables

What machine learning algorithm would be the most optimal to use for a classification problem with a small number of classes? In my case, only two. The sample size is also rather small (<100), but ...
0
votes
0answers
27 views

Advantages of LS-SVM over SVM

My teacher asked me to do a research on LS-SVM, I know what is LS-SVM and how its mathematically different from SVM. I have found lots of papers that shows that for large-scale problems LS-SVM have ...
0
votes
0answers
16 views

model choice multi state sequence prediction (and possible R packages for solving problem)

I have a set of sequences (dataset) where I have sequences of letters. I also have a corresponding response sequences where the known state of the sequence are. I would very much like to make a ...
0
votes
0answers
24 views

Converge in Probability of random variables

I don't know if I understood the (Convergence in probability of random variables) formula, why $ | Xn -X | $ should be $$ \geq \varepsilon $$ For example if $\varepsilon $=5( a random number), and ...
4
votes
2answers
76 views

The right way to use Machine Learning to predict latitude and longitude

There are some simple ML techniques that can be used to easily predict latitude/longitude co-ordinates, such as predicting the latitude and longitude separately using two different models. However, I ...
1
vote
1answer
16 views

How to map data to another feature space

I have some data which is described in a feature space $F$. Let's call this dataset $X_F$. That is, $X_F$ is a matrix where each row an instance and each column is a feature (characteristic). Suppose ...
0
votes
1answer
53 views

Python machine learning brute force [closed]

I have a dataset from which I try to predict one value. I tried linear regression, but it doesn't work. Is there an algorithm in Python which will compute a good statistical model (neural network) in ...
1
vote
0answers
21 views

Tuning priors/weights/costs to counteract class imbalance

I have a classification problem which consists of two classes. It has high class imbalance. There are around 85% data points for the negative class and only 15% for the positive class. One option is ...
0
votes
0answers
12 views

classification for standard normal features

I have an artificial binary classification problem and I know each feature follows a standard normal distribution. For example, we have some standard normal independent features $x_1,x_2,...,x_n$, I ...
0
votes
0answers
30 views

eval_set on XGBClassifier

can someone explain what does the eval_set parameter do on the XGBClassifier? I thought that by using eval_set, the algorithm would do some sort of grid search and find the best model to fit on train ...
2
votes
1answer
87 views

Getting hints from target variable: Will it ruin predictive power completely?

I have a large set of predictors and a target variable which is extremely difficult to model. After a couple of failed trials (glm, DT, RF, NN) I got the impression that it is almost random noise. ...
0
votes
0answers
6 views

Inter annotator aggrement for unbalanced datasets - KAPPA statistics

Can you suggest a better Kappa to measure the agreement between two annotators for unbalanced datasets with two classes?
1
vote
0answers
26 views

Unsupervised Aspect based opinion mining [closed]

I have been working on a project to create a model for Unsupervised Aspect based opinion mining on online reviews ( broadly domain independent). i want to automatically extract features of the ...
0
votes
1answer
14 views

Cross validation error dependency?

Let's say we are running CV with K folds. Can you give an intuitive explanation on why the errors per fold are dependent? I was asked this and after thinking about it I kind of see it but need some ...
0
votes
1answer
20 views

What advices do you have for a starter in multiple image recognition?

So, I have experience in machine learning for NLP and a little in neural networks for NLP, but never so far done anything in computer vision in this area so bear with me if what I am asking is a ...
1
vote
1answer
31 views

Given a prediction problem, what principles drive the design of a neural network for that problem?

Research work that have solved problems using neural networks, simply state the structure of their networks in their research papers. No explanation is generally given about what led them to that ...
2
votes
1answer
20 views

How to design a classifier when new data has missing values

I have trained a classifier for medical data, which works ok. Now I have to build a final product to give to the MDs (a sort of program where you give a new patient's record as input and the ...
0
votes
1answer
27 views

label of non-classified examples in machine learning

What happens if a positive example in neither classified as positive nor negative? Will you treat such an example as False Negative (FN)? I am using a tool (machine learning) that predicts an ...
1
vote
1answer
45 views

Neural Network: For Binary Classification use 1 or 2 output neurons?

Assume I want to do binary classification (something belongs to class A or class B). There are some possibilities to do this in the output layer of a neural network: Use 1 output node. Output 0 ...
0
votes
0answers
8 views

Frame work for neural net and recurrent net

I would like to run neural net and recurrent net (such as jordan net and elman net and so on) program using java program language. Is there anyone who recommend me to useful flame work? Thank you in ...
4
votes
1answer
55 views

How was this intergral derived from Bayes' Rule in David Heckerman's Bayesian Network paper?

I am trying to follow this paper titled "A Tutorial on Learning With Bayesian Networks" by Microsoft researcher David Heckerman. In it I am unable to figure out how he got to Equation 2 from ...
1
vote
1answer
34 views

How do I know when a Q-learning algorithm converges?

I am currently trying to implement the Q-learning algorithm. After reading enough to have a good understanding of how it works, I am now wondering how to know when the algorithm actually reaches ...
2
votes
1answer
52 views

Deep Learning on a weird dataset [closed]

I'm currently learning Deep learning and I'm trying to create a dnn on a dataset with difficult data to learn. There is a lot of continous attributes and there seem to be little to no difference ...
1
vote
0answers
57 views

Linear SVM feature weights interpretation. Binary classification, only positive feature values

I'm using clf = svm.SVC(kernel='linear') on a data set with only two classes $y \in \{-1, +1\}$ and the feature values of all samples are normalized between 0 and ...
1
vote
1answer
72 views

Difference between ElasticNet in scikit-learn Python and Glmnet in R

Has anybody tried to verify whether fitting an Elastic Net model with ElasticNet in scikit-learn in Python and glmnet in R on ...
0
votes
0answers
42 views

Gradient descent applied to softmax regression

I am currently trying to reimplement a softmax regression to classify MNIST handwritten digits. I not a machine learner and my plan was to get an intuition of the entire workflow that has to be ...
1
vote
0answers
41 views

Classifying dataset with few positive results, how to handle false positive paradox?

I'm working on a classification task with data that have only 2% of positive results and it's hard to get high true positive result with low false positive. Maybe you know how to handle such data, or ...
0
votes
0answers
27 views

Residual network dimension changing blocks identity function

In trying to implement ResNet with bottleneck blocks for myself, I got very confused about the identity function residual blocks with different dimensions. They compared identity, conv projections on ...
0
votes
0answers
12 views

Combining results from tests after re-shuffling data

I am fine-tuning a neural network for a binary multilabel classification problem. Basically I am trying to predict 64 binary labels for each input. However, my dataset is somewhat small for the task ...
1
vote
0answers
38 views

Server Requirements for machine learning on 300 million row database [closed]

I am trying to build a logistic regression model using Revolution R 64 bit. My data is 300 million rows by 12 columns. The data is stored on AWS Redshift, but takes about 6 hrs to import and save as ...
0
votes
0answers
35 views

Assumptions behind cross-validation

According to "no free lunch theorem" (also here and here), we cannot deduce just from the data alone (without any domain knowledge) which classifier is better. Of course, we use cross-validation to do ...
1
vote
0answers
11 views

Question regarding Gaussian Discriminant Analysis, and Generative Learning models

In lecture today, my professor mentioned in the context of GDA and Generative learning, we would like to learn the joint probability $P(x, y)$, where $x \in \mathbb{R}^n$ and $y \in \{+1, -1\}$. ...
0
votes
0answers
13 views

Pattern recogniton in measured data

I am looking for a way to find patterns in a large amount of data and compare it to a reference set. To give an example: I have as reference the data about power consumption of a machine during the ...
0
votes
0answers
24 views

Learning matrix with gaussian process

Assume we have a matrix $A$ and that its rows are normally distributed (we assume a gaussian prior for the rows of A). Now, we want to learn the matrix A. The problem I find is in determining the mean ...
1
vote
0answers
10 views

Detecting Frequency in Noisy Data

I have some very noisy data that seems like it might have a frequency to it. I'm trying to build a model with the data, like the example code below. So I've been experimenting with fourier series ...
1
vote
0answers
30 views

Sparsity on the simplex

Say I want to minimize a convex function on the probability simplex. How is it possible to encourage the sparsity of the solution (while keeping the problem convex)? Since using the sparsity ...
0
votes
0answers
69 views

Neural Network for Reverberation Decay Rate Prediction (Keras)

I'm not sure if this is the right place to post this question, sorry if it's not. Essentialy I’m trying to use a neural network to predict decay rate from a reverberant audio sample. I have taken a ...