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

learn more… | top users | synonyms (1)

0
votes
0answers
14 views

cross-validation: what is the standard deviation if the same value is obtained for each fold?

Here is a detailed imaginary example: I am using 5-fold cross-validation to estimate the generalization MSE of my predictive model. When I hold-out fold number 1, which contains 10 observations, say ...
0
votes
0answers
13 views

Cost function spiking upon using dropout on neural network

Upon using the dropout technique, my cost function is spiking arbitrarily. Is this normal? If not, how do I avoid it? I'm using a salt-and-pepper mask to drop out neurons at a dropout rate of 5%. ...
0
votes
1answer
25 views

Lagrangian multiplier: role of the constraint sign

I am beginner learning Lagrange multipliers with wiki article. Consider: maximize $f(x,y)$ subject to $g(x,y) = 0$ I understand that to maximize I must follow the gradient $\nabla {_{x, y}}^{}f$. I ...
0
votes
0answers
19 views

How do I use MatLab or R data model in an machine learning application [on hold]

I want to use matlab or R to create a prediction model, once this is complete how would I use that model in a an application?
0
votes
1answer
30 views

Classification problem where one attribute is a vector

Hello I am a layman trying to analyze game data from League of Legends, specifically looking at predicting the win rate for a given champion given an item build. Outline A player can own up to 6 ...
1
vote
0answers
12 views

Is there a classification model with possibility to set transition probabilities for each sample?

There are models where we can to set initial transition probabilities between classes like Markov models. But what if we need to change the probabilities dynamically depends on all previous ...
0
votes
1answer
15 views

Preprocessing via PCA in Caret, then fitting PLS

I am dealing with quite highly-dimensional data, and am using (in R) Caret's preprocessing 'pca' method to reduce the dimensionality. However, dependent on the number of components I choose, I seem to ...
0
votes
1answer
21 views

Composite similarity of GitHub projects

As a hobby, I'm building recommendation system which finds related projects on GitHub. I'm computing Jaccard index for each repository, based on users who gave stars: $$J(A, B) = \frac{ \left\vert ...
0
votes
0answers
18 views

Why is my simple implementation of sub-gradient descent for SVM not converging?

As an exercise in understanding the mechanics of the Support Vector Machine, I am attempting to implement the SVM myself, in Python. I'm more concerned with understanding than efficiency, so I wish to ...
0
votes
1answer
23 views

Sampling : Gradient Boosting Tree

I have a question regarding the algorithm of Gradient Boosting Tree. I understand Simple tree is built for only a randomly selected sub sample of the full data set (random without replacement). Each ...
0
votes
1answer
28 views

What is the difference between kalman filter and extended kalman filter?

I am working SLAM based problems in robotics and I want to know whether I can use Kalman filter instead of the Extended kalman filter that is predominantly used ? If not, what is the difference?
0
votes
0answers
8 views

Multiple Response Regression in Spark MLLib

I am trying to do a regression using RandomForests in Spark ML where I have several input variables and would like to predict several responses. Training data would look like X = ...
0
votes
0answers
5 views

Equivalent method to simplex algorithm in machine learning

I always use simplex algorithm for minimization problems. Is there an equivalent approach in machine learning that could possibly be better, smarter?
0
votes
0answers
12 views

Multi-tier classification

First of all, I'm not sure wether the question title is correct, but I'm facing a puzzling problem. Please point me to the correct term and some relevant literature. This is the problem: Let's say I ...
0
votes
0answers
22 views

sentiment analysis using convolutional neural networks

I was trying to modify YoonKim's code for sentiment analysis using CNN's. He applies three filters of heights=[3,4,5] and ...
10
votes
4answers
772 views

Why splitting the data into the training and testing set is not enough

I know that in order to access the performance of the classifier I have to split the data into training/test set. But reading this: When evaluating different settings (“hyperparameters”) for ...
1
vote
0answers
52 views

Comprehensive list of misnomers in machine learning

Are there any reference document(s) that give a comprehensive list of misnomers in machine learning? I would like to have a list and simple explanation if needs be that I could go through easily (vs. ...
0
votes
0answers
14 views

Predicting lat/long from binary features

I have a number of observations that occur around my city (a small area), and several of them have latitude and longitude. I have been looking into predicting the latitude/longitude of the ...
0
votes
0answers
10 views

Non linear dimension reduction method that provide a way to map data from embedded space to original data space

I am looking for dimension reduction tools that allow to map data from the low dimensionnal embedded space to the high dimensionnal space of the original data. For example, Gaussian process latent ...
2
votes
0answers
28 views

Boosting: why is the learning rate considered a regularization parameter?

I understand that the learning rate parameter ($\nu \in [0,1]$) in Gradient Boosting shrinks the contribution of each new base model -typically small trees- that is added in the series. It was shown ...
0
votes
0answers
13 views

What algorithm to use in order to find pairs of numbers belonging to one person? [on hold]

There are data. In each row: Lac---ID of the mobile stations' group cid --- station' ID within LAC msisdn--phone number imei--device number. Based on the first 8 numbers: ...
0
votes
0answers
26 views

How to interpret prediction accuracy based on error analysis in test and validation sets?

I would like to calculate accuracy of my prediction model based on polynomial regression and I got some values for the test and validation errors 3.895 and 4.0125. I am wondering how these values for ...
0
votes
1answer
37 views

Machine Learning: What are the types of data set?

Is there a classification of the type of data sets in machine learning problems? I am specifically interested in classification problems. I know a large number of algorithms like SVM, NN, Decision ...
12
votes
4answers
1k views

Why is multicollinearity not checked in modern statistics/machine learning

In traditional statistics, while building a model, we check for multicollinearity using methods such as estimates of the variable inflation factor (VIF). But in machine learning, we instead use ...
0
votes
0answers
32 views

Pros&Cons of Hidden Markov Models in Time Series Forecasting

What are the advantages and disadvantages of Hidden Markov Models in forecasting values of a time series (compared to other methods, e.g. ARIMA)?
0
votes
0answers
14 views

Anomaly Detection: Pattern Recognition inTime Series

I am trying to implement an anomaly detection tool based on Pattern Recognition. The data I am working on are periodic.I extract the pattern from a training set and then compare it to data in the ...
1
vote
1answer
35 views

verified procedure for calculating gradient descent?

I'm wondering if there is a good procedure people are using for gradient descent that is pretty well validated--something like a package for R or Python, or generic code many people adapt. After ...
1
vote
0answers
14 views

What optimization methods work best for LSTMs?

I've been using theano to experiment with LSTMs, and was wondering what optimization methods (SGD, Adagrad, Adadelta, RMSprop, Adam, etc) work best for LSTMs? Are there any research papers on this ...
0
votes
0answers
15 views

Existing relationship between predictors and response in a classification method

Lets say that there are 3 predictors A,B and C. There is a response Y, which is already related to A,B and C mathematically. Y = ABC. Past data for Y does not exist, and is calculated using the above ...
0
votes
0answers
10 views

How to add extra layer of MLP to DBN

I am trying to add MLP layer to DBN that can use final parameters of DBN model as Input for MLP model. I am new to python so am not well versed with its input and output processes. Any help is ...
0
votes
1answer
32 views

Why is a Euclidian metric better than a simple one for this ML problem?

I'm trying to learn ML, and have been reading Machine Learning Projects for .NET Developers by Mathias Brandewinder (I'm a C# developer by day, so this appealed to me). He uses the digit data ...
1
vote
1answer
61 views

Role of probability in machine learning

In any machine learning example say linear regression, the objective is sometimes two-fold. One to optimize some analytical function like squared error and one is to optimize some probability ...
1
vote
1answer
26 views

Most Informative Features with Naive Bayes

Anyone know how to calculate the most informative features where the attributes are normally distributed using Naive Bayes? My understanding, at least if you have binary attributes, is that you ...
1
vote
1answer
33 views

Does removing mildly correlated features (0.5) improve performance in predictive models? (SVM, random forests)

I am trying to model a binary response using a 500+ dataset. I already removed many non useful features in order to reduce dimensionality and improve my model. I am wondering whether in general ...
0
votes
1answer
20 views

Which parameters to tune in CART?

I am using caret package in R to train CART model. train function seems to tune only the complexity parameter (which in a way determines depth of the tree and number of terminal nodes). Is this ...
1
vote
0answers
14 views

The quality of a training corpus

When we train a classifier/regressor we need a training corpus, a general recollection of items pre-classified. Our best classifier is good as the corpus used to train it. This means that given a ...
0
votes
0answers
10 views

Is a test set sampled after training examples have been removed by an active learning algorithm valid?

In this case desire is to optionally label a final test set after training. The training is to be done using an active learning approach which can be biased towards different characteristics ...
0
votes
0answers
10 views

calibration of model using true positive rates for low frequency classess

I am conducting a classification problem on a very low frequency class. For example there just 20 samples in 1,000 samples which should be classified from other. I get very high accuracy but not ...
0
votes
0answers
41 views

Causal Trees to Estimate Heterogenous Treatment Effects: Transformed Outcomes [Machine Learning in Python]

I am interested in using off-the-shelf tools like scikit-learn for Python to implement the Athey-Imbens recommendation for estimating treatment effect ...
0
votes
0answers
52 views

Research in high-dimensional statistics vs. machine learning? [closed]

As a PhD student starting to think about dissertation topics, I am particularly interested in high-dimensional statistical learning. I wish to find some research review/survey/papers (or webpages, ...
0
votes
0answers
15 views

error family in boosted regression tree: gbm package

I am trying to understand boosted regression tree. I am using the gbm package in R. I noticed that in this package one has to specify the error family. I understand ...
0
votes
1answer
47 views

(Metric?) Machine Learning?

I have a set of data objects defined by 20 dimensions rated from 1 to 10 with no decimal. There is no hierarchy between dimensions. I am able to calculate the similarity between objects but I do not ...
6
votes
2answers
386 views

Why do Convolutional Neural Networks not use a Support Vector Machine to classify?

In recent years, Convolutional Neural Networks (CNNs) have become the state-of-the-art for object recognition in computer vision. Typically, a CNN consists of several convolutional layers, followed by ...
0
votes
0answers
26 views

Deep learning how to study? [closed]

What is the best way to study deep learning? Is there a website that can I find self learning material?
3
votes
3answers
100 views

Kernel SVM: I want an intuitive understanding of mapping to a higher-dimensional feature space, and how this makes linear separation possible

I am trying to understand the intuition behind kernel SVM's. Now, I understand how linear SVM's work, whereby a decision line is made which splits the data as best it can. I also understand the ...
0
votes
1answer
17 views

Evaluation of binary approach to one vs all multi-class classification

I'm working on a multi-class problem which I have redefined as a series of binary problems (i.e. a one vs all classification problem). However, each observation can belong to more than one class. For ...
2
votes
0answers
41 views

Stories about statistics, machine learning and data mining [closed]

I'm looking for interesting stories about statistics, machine learning, etc. Examples of stories I find interesting are the following: https://en.wikipedia.org/wiki/Carl_Friedrich_Gauss developed the ...
1
vote
0answers
16 views

How to use RFECV for feature selection and cross validation

I am still very new to machine learning and trying to figure things out myself. I am using SciKit learn and have a data set of tweets with around 20,000 features (n_features=20,000). So far I achieved ...
2
votes
1answer
17 views

G-Means: Learning K in K Means

I am currently trying to understand the paper 'Learning K in K-Means' by Hamerly & Elkan (2004), which is the paper implementing G-Means, an automated way in selecting K for K-Means. There is one ...
0
votes
0answers
7 views

Understanding divisive clustering code

I came across divisive clustering algorithm (Cichosz, P. (2015) Data mining algorithms: Explained using R; page 262), which is implemented in R. The appropriate function is pasted below. Actually, I ...