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

learn more… | top users | synonyms

6
votes
4answers
314 views

The daily job routine of the machine learning scientist?

I'm a master CS student in a German university now writing my thesis. I will be done in 2 months I have to make the very hard decision if I should continue with a PhD or find a job in the industry. ...
2
votes
0answers
18 views

Training a Tic Tac Toe brain - am I on the right track?

My only experience with Machine Learning is Andrew Ng's Coursera course, but I did work through that just fine and passed with 100%. I decided to practice by making up some problems and solving them. ...
0
votes
1answer
13 views

Is Area under curve a composite function

I have some data examples. If I split the data into three parts and the have some scores for each example of the three parts and then calculate individual AUCs for the three parts In the next case, I ...
0
votes
0answers
7 views

Updating SVD in Recommender Systems for change in ratings

I have read that there are projection based methods to accomodate for new user's ratings or for the ratings for a new item in SVD. However, I want to know how to update my feature space for change in ...
0
votes
0answers
3 views

how to implement linear or non linear regression for 3d position estimation in Matlab? [on hold]

I am a beginner in Machine Learning. For my project I need a regression algorithm that can estimate the 3D position of a device based on some constrains (more over inputs). I know how to implement ...
7
votes
1answer
103 views
+50

Bayesian lasso vs ordinary lasso

Different implementation software are available for lasso. I know a lot discussed about bayesian approach vs frequentist approach in different forums. My question is very specific to lasso - What are ...
6
votes
1answer
55 views

How to decide which penalty measure to use ? any general guidelines or thumb rules out of textbook

A number of regularization measures are available in literatures, which is kind of confusing to beginners. The classical penalty is ridge by Hoerl & Kennard (1970,Technometrics 12, 55–67). ...
0
votes
1answer
16 views

Regression-tree Tuning in a Streaming Setting

Some time ago I went through a NIPS 2013 paper Regression-tree Tuning in a Streaming Setting. The paper proposes a tree-based regressor. Is there any implementation of this algorithm available? (At ...
1
vote
1answer
39 views

Statistical testing: Multiple classifiers, 1 domain. Would rANOVA be appropriate?

When comparing the performance of two classifiers over a single domain, in the context of a classification problem in machine learning, it is common to use a paired t-test, using the 10 average ...
1
vote
1answer
45 views

For a model like this what performance measures can I calculate and how?

Methods: From the machine learning literature, I understand different parameters can show performance of model in machine learning. I would briefly expand my understanding with confusion matrix: ...
2
votes
1answer
31 views

What's the optimal way to encode a 'month' feature?

What's the optimal way to encode a 'month' feature? A single integer value or 12 binary values don't quite grasp the concept of modulo distance... Say I want to train an SVM for a certain task and ...
2
votes
2answers
52 views

Guideline to select the hyperparameters in Deep Learning

I'm looking for a paper that could help in giving a guideline on how to choose the hyperparameters of a deep architecture, like stacked auto-encoders or deep believe networks. There are a lot of ...
0
votes
0answers
21 views

10 fold cross validation model in weka

down vote favorite I am new to Weka. Trying to build a specific Neural Network arcitecture and testing it using 10 fold cross validation of a dataset. Now building the model is a tedious job and ...
0
votes
0answers
13 views

Number of Predictors and Classification Algorithm

In general, is it better to include more predictors in algorithms such as SVMs and random forests compared to logistic regression? It seems that when we add more predictors to logistic regression, the ...
3
votes
1answer
155 views

Prediction of continuous variable using “bnlearn” package in R

I use bnlearn package in R to learn the structure of my Bayesian Network and its parameters. What I want to do is to "predict" the value of a node given the value of other nodes as evidence ...
4
votes
1answer
133 views

Reinforcement learning of a policy for multiple actors in large state spaces

I have a real-time domain where I need to assign an action to N actors involving moving one of O objects to one of L locations. At each time step, I'm given a reward R, indicating the overall success ...
2
votes
1answer
21 views

What do NORB and CIFAR stand for?

The MNIST dataset is a standard benchmark data set of digit images. MNIST stands for 'Mixed National Institute of Standards and Technology'. The NORB dataset is a commonly used dataset of binocular ...
3
votes
3answers
41 views

In Naive Bayes, why bother with Laplacian smoothing when we have unknown words in the test set?

I was reading over Naive Bayes Classification today. I read, under the heading of Parameter Estimation with add 1 smoothing: "Let $c$ refer to a class (such as Positive or Negative), and let $w$ ...
1
vote
0answers
13 views

Difference between Factorization machines and Matrix Factorization?

I came across the term Factorization Machines in recommender systems. I know what Matrix Factorization is for recommender systems but never heard of Factorization Machines. So what's the difference?
2
votes
3answers
217 views

How to combine time-series based features with different frequencies

I have 3 features which I want to use in my classifier. They are all time-series data-based. However, they are all at different frequencies and there have different matrix dimensions. I was wondering ...
1
vote
0answers
10 views

How to assess the importance of the features which come from intersection of features of the two models?

I have two models from two different data sets. Model 1 contain 50 features and model 2 contain 40 features. the intersection of features of model 1 and 2 is 10. so how can I assess the relative ...
0
votes
1answer
113 views

Machine learning with trinomial features

I have 100,000 students who have each answered some multiple choice questions. Given their performance I want to work out what the chances are of a particular student answering the next question ...
1
vote
2answers
112 views

High precision with low recall SVM

I'm classifying a data set using SVM and those are the precision and recall values for two classes. ...
1
vote
1answer
52 views

Linearly dependent features

I have a matrix A of 1000 observations (rows) and 100 features (cols). I would like to find: Linearly dependent features so that I can remove them and simplify the problem. rank(A) gives me 88, ...
-2
votes
0answers
13 views

Error in eval(expr, envir, enclos) : object 'Case' not found in train( method=rpart) in caret package of R [on hold]

I am trying to fit my training data for Coursera Practical Machine Learning Quiz 2. ...
1
vote
1answer
17 views

SVM Classification with Duplicate Training Instances

I'm using SVMs with linear kernel for sentence classification (binary). My dataset contains many duplicate instances i.e. many sentences in the training set have identical feature vectors. In the ...
0
votes
1answer
43 views

What are the general strategies in creating a Probabilistic Graphical Model?

While there is lot of theory and probability in the background to understand, I wanted to know if there are any resources/quick pointers on what to consider while modeling a problem using Bayesian ...
0
votes
1answer
12 views

How to compare the nested models which each of them comes from diffrent dataset?

I have four nested models.Every of them learned from different data sets. now I want to compare these models together.normally people try to compute the F-satistics. But for my case, it's bit harder, ...
9
votes
1answer
330 views

Can a model of P(Y|X) be trained via stochastic gradient descent from non-i.i.d. samples of P(X) and i.i.d. samples of P(Y|X)?

When training a parameterized model (e.g. to maximize likelihood) via stochastic gradient descent on some data set, it is commonly assumed that the training samples are drawn i.i.d. from the training ...
0
votes
1answer
23 views

k-means clustering on percentages

Can we do k-means clustering on percentage data (like 56%, 44%, 22%, 13%, etc.)? There is a data set, and data in various parts are measured in percentages.
0
votes
0answers
12 views

Expected required sample length to train a hidden Markov model

Say one wishes to train a hidden Markov model with $n$ hidden states, and (accidentally) the problem itself can be described with a hidden Markov model with $n$ (or less states). What is the expected ...
0
votes
1answer
13 views

Assumption behind few latent features in recommender systems?

I know in recommender systems you have a rating matrix and then you factorize this matrix into two matrices and then learn those matrices with gradient descent. In those matrices we specify the number ...
0
votes
1answer
32 views

Approach for mapping consumer preferences

I have this web application where I need to map consumer preferences based on some input information and individual choices. My goal is to create a list of product recommendations and evaluate the ...
2
votes
1answer
27 views

Multi-armed bandit in face of full reward information

I am new to this area of machine learning. I am just walking myself through UCB1 algorithm which seems to assume that the payoff can be learnt only for action that ...
1
vote
1answer
61 views

How do you validate your machine learning models?

I am wondering what approaches are commonly used for validating a classification or prediction models: Approaches that am using at the moment: Using truth-sets: - ROCs, Bootstrapping, Accuracy, ...
0
votes
0answers
15 views

Forensics in wireless networks, anomaly detection and beyond?

first i'de like to apologize if this is not the right place. Next year i'm gonna be working on my final project in computer security, i have to build a wireless forensics tool that can analyse a data ...
1
vote
1answer
21 views

Feature Selection - Mutual Information with response variable that takes three values

I am trying to calculate Mutual Information scores for Feature Selection. I have successfully implemented the Mutual Information to test each feature against the binary response variable. Each ...
0
votes
1answer
25 views

Interpretation C value in linear SVM

My C value is very low (close to 0). Does this mean that my feature (dimensions) have no real separative (and thus predictive) value? (As the SVM basically chooses to ignore the training data ...
1
vote
1answer
49 views

Simple SVM Question

For a linear SVM, the documentation tells me the formula is: $$ \frac{1}{2}w^Tw+C\sum\limits_{i=1}^l\xi_i$$ Please explain to me in layman's terms what w (and ξ) represent. Is w the distance to the ...
0
votes
0answers
28 views

large variables and low sample (p > n) problem: ridge , LASSO, PLS, PCR which is most suitable for predictions

I am trying see whether to go for ridge regression, LASSO or principal component regression (PCR) or Partial Least Squares (PLS) in a situation where there are large number of variables / features (p) ...
1
vote
0answers
13 views

Should the 'TPR of adaboost' be better than base classifiers'?

There are two different base classifiers which produce true positive rate (TPR) values 99.46% and 91.79%. When I use these base classifiers in adaboost, what should the new TPR be? Better than two of ...
4
votes
1answer
135 views

Regarding the sampling procedure in Adaboost algorithm

The AdaBoost algorithm states that it is to train a classifier based on the training data according to a weight vector. Assume the size of training data is N, the weight vector is of dimension N as ...
0
votes
1answer
53 views

error increasing with no of estimators in adaboost

My error gets increased when i increase the n_estimators value in ...
1
vote
1answer
40 views

How does cross-validation and the Bayesian method overcome the overfitting problem?

I was told that cross-validation and the Bayesian method can overcome the overfitting problem. I was told that comparison of models in a Bayesian way is actually doing cross-validation... What is the ...
2
votes
1answer
49 views

How to understand the log marginal likelihood of a Gaussian Process?

I'm trying to understand Gaussian Processes. Could anyone tell me: Why we need to use the log marginal likelihood? Why using log, the marginal likelihood can be decomposed to 3 terms (including a ...
2
votes
1answer
148 views

Comparison of two classifiers based on precision/recall/F1 only?

For two classifiers h1 and h2 I have the precision, recall and F1 score as a percentage (along with the original labeled data set that they were tested on). If I had access to which samples each ...
1
vote
1answer
104 views

Machine-learning input data distribution

I'm trying to build a binary 1/0 ML classification algorithm, and was thinking about how to set up the input dataset. If the event I want to predict (the 1's) occur relatively less frequently in the ...
1
vote
1answer
531 views

Decision boundaries and Gaussian density functions

This is for my hw, and if anyone can solve the first part of the question it will be great. Here is the question: Assume a two-class problem with equal a priori class probabilities and Gaussian ...
0
votes
3answers
78 views

What is shallow architecture in machine learning?

What is a precise definition of shallow architecture in machine learning?
1
vote
2answers
69 views

Class labels in data partitions

Suppose that one partitions the data to training/validation/test sets for further application of some classification algorithm, and it happens that training set doesn't contain all class labels that ...