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

learn more… | top users | synonyms (1)

3
votes
0answers
93 views

Which, if any, machine learning algorithms are accepted as being a good tradeoff between explainability and prediction? [on hold]

Machine learning texts describing algorithms such as gradient boosting machines or neural networks often comment that these models are good at prediction, but this comes at the price of a loss of ...
1
vote
1answer
63 views

what is “principled feature selection”?

i see the expression "principled feature selection" in titles of various Machine Learning papers and generally in the literature but nowhere do authors really define what they mean. "principled" as ...
0
votes
1answer
15 views

Baum Welch:Calculating transition probability

I am trying to understand the Baum Welch algorithm by implementing it in xls. I have chosen a simple example of observations from a loaded (L) vs fair (F) die. I calculate the forward and backward ...
0
votes
0answers
15 views

Confusion over the test error and the expected error and actual implementation

Given a training set, $\tau = \{(x_1,y_1,\dots, x_N,y_N \}$ and a model $\hat{f}(x)$ has been fit. We have the following two definitions: The Generalisation (Test) Error $$ Err(\tau) = ...
0
votes
1answer
18 views

How do GBR trees differ from random forests regression in terms of predictive performance?

Is there a case when one would use gradient boosted regression trees instead of random forests regression (or vice versa)? It appears gradient boosted regression trees have done far better in ...
1
vote
0answers
42 views

Which classifier to use? [closed]

I have a course project that I need to finish. I'm using Weka 3.8 and I need to classify text. The result needs to be as accurate as possible. We received a train and a test .arff file. We need to ...
1
vote
1answer
42 views

Deriving the intercept term in a linearly separable and soft-margin SVM

I have read Andrew Ng lecture notes on Support Vector Machines as well as the notes from MIT OpenCourseWare and I have a few doubts concerning the derivation of the intercept value: First, there is ...
1
vote
0answers
17 views

How to sample weights for weighted kernels?

I'm using a SVM classifier with a weighted RBF kernel. My dataset has 17 features. In the RBF kernel I will use a weight for each feature. Of course the weights must sum to one. For choosing the best ...
1
vote
0answers
35 views

How does one infer the noise/error model given measurements?

What resources are available for applying inference/computational statistics to infer the underlying error/noise model, given X measurements from some apparatus? (Below, I am mostly referring to ...
0
votes
0answers
17 views

Parameter ranges for sigmoid and polynomial kernel

I would like to use a SVM classifier with a sigmoid and polynomial kernel. The sigmoid kernel has the following form: $K(u,v) = \tanh(\gamma * u'v + \text{coef}_{0})$ The polynomial kernel has the ...
0
votes
0answers
78 views

Discriminant function of 1 Nearest Neighbor

Consider the following question: We will consider the case of 1-nearest neighbor, and look at the details of computing the error probability. In this case, let us assume that we have two classes, ...
0
votes
0answers
12 views

'undirected' $\epsilon$-greedy action selection

There are two famous/classical ways to select an action under the $\epsilon$-greedy action selection that discusses the trade off between exploration and exploitation. Firs is the Semi-uniform ...
1
vote
0answers
15 views

How to make sense of the EM algorithm expressed in terms of Kullback-Leibler divergence?

In the textbook by Wasserman, All of Statistics, the Algorithm is expressed as: 1) Pick a starting value $\theta^0$. 2) (E-Step). Calculuate: $$ J(\theta|\theta^j) = E_{\theta^j} \left(log ...
2
votes
2answers
20 views

Does a linear classifier has spatial awareness?

Say we are trying to classify images using a linear classifier, and in our training set we have say cars in the middle on a white background. If in our test set, we shift the cars to the right but ...
0
votes
1answer
24 views

Comparison of LDA vs KNN time complexity

Which algorithm has a better performance in terms of time complexity, LDA or KNN?
0
votes
2answers
37 views

Is feature engineering relevant at all for Random Forests?

Random forests is an ensemble of trees that learns the hidden patterns in the data. I have mostly tried doing some feature-engineering before running the Random Forest model but is it required or the ...
0
votes
0answers
20 views

WEKA / Support vector machine - does attribute have positive or negative impact?

I'm doing some machine learning with WEKA using around 20 numeric attributes to predict a true/false class. Support vector machine gives me good enough results on my data set. However, I'd also like ...
1
vote
1answer
29 views

Parameter selection and k-fold cross validation

I have one dataset, and need to do cross-validation, for example, a 10-fold cross-validation, on the entire dataset. I would like to use radial basis function (RBF) kernel with parameter selection ...
0
votes
1answer
34 views

Multi-class SVM Calibration

Say we have multiple SVMs used in a one-vs-all approach, such that classes a, b, c correspond to 3 SVMs trained positively on the class and then negatively on all ...
0
votes
0answers
10 views

Is F-score (precision-recall) meaningful for a balanced dataset?

In the binary classification scenario, for highly unbalanced datasets, when the task is to find a few needles in a haystack -- say, in information retrieval -- we use F-score (precision-recall) for ...
1
vote
2answers
30 views

Is Naive Bayes suitable for large datasets with thousands of features?

I have a data set with 100 million rows and 15,000 categorical variables each with 0/1 values. My target variable is also a 0/1 binary variable. Is Naive Bayes suitable in terms of computational ...
0
votes
0answers
9 views

How to design a fitness function for binary logic network?

Assume we have a directed graph of connected nodes, where each node represents logical operator. Input for this logic operator are values on all edges leading to the node and result is outputted to ...
0
votes
0answers
11 views

What are Linearly Solvable MDPs?

Markov Decision Process (MDP) is a formalism mainly used in artificial intelligence on the structure of decision making of a learner/agent. The aim is to find a suitable policy that maximizes the ...
0
votes
1answer
20 views

How do I choose the initial features vectors for a Stochastic Gradient Descent trained SVD++ algorithm?

I'm reading the SVD++ Netflix Recommender Systems paper because I want to be able to properly assess this approach to building a recommender system. How should I choose the initial values of $q_i$ ...
0
votes
0answers
19 views

Predicting value with MLP neural network

I am trying to predict a value using feedforward and back-propagation network but I get incorrect predicted value. I checked the gradients calculation and error is 1.79684799923e-11 < 1e-9 so ...
0
votes
0answers
19 views

Need new strategy for single class classifier

I am attempting to create a single class classifier where the classes are fairly close to balanced (+/- 25). My dataset has ~2,800 samples and ~1,100 features. All of the features are binary except ...
1
vote
0answers
86 views

Data structure for rare event predictions in temporal domains

I am a beginner in rare event modeling. I am working on predicting modem failures within a network where failures occur approximately 3% of the time. Currently my data is structured as follows: ...
0
votes
3answers
40 views

Multiclass classification question

I am working on applying Random Forests to a multiclass classification problem, where I have a set of 11 predictor variables and a response that can take the values of "Yes", "No", and "Maybe". In my ...
0
votes
0answers
9 views

What is a channel when referring to a convolutional neural network?

I know that an image can have multiple channels (e.g. red, blue and green channels). I'm reading a paper where each convolutional neural network layer has 24 channels with a 3x3 kernel. What does ...
0
votes
2answers
70 views

How to choose between ROC AUC and F1 score?

I recently completed a Kaggle competition in which roc auc score was used as per competition requirement. Before this project, I normally used f1 score as the metric to measure model performance. ...
0
votes
0answers
29 views

Why getting better classification results despite many irrelevant terms?

I am new to ML especially for document (text) classification. I have 22 classes (scientific fields) and I am trying to improve classification results by employing some additional data. That is, I use ...
0
votes
0answers
32 views

General approach to learning a graphical model

Lately I've been reading a lot about inference and learning in probabilistic graphical models. I mostly understand specific methods (e.g. junction tree, message passing, MCMC; gradient descent, ...
15
votes
1answer
760 views

What are some good interview questions for statistical algorithm developer candidates?

I'm interviewing people for a position of algorithm developer/researcher in a statistics/machine learning/data mining context. I'm looking for questions to ask to determine, specifically, a ...
1
vote
0answers
11 views

How to mitigate the hierarchical error propagation in tree-structured classification

Suppose we have a multi-class classification problem, where the number of classes $K \geq 3$ We use a tree structure of multiple SVMs to divide and conquer the problem, with one example in the figure ...
2
votes
0answers
47 views

Interpret ROC/AUC values with respect data

I am using R to plot ROC curves. I have a prediction matrix, where each column shows the prediction values corresponding to different approaches. Also, I have a ...
0
votes
1answer
12 views

taking average of several models and feature sets

just a quick question that i cant seem to find a definitive answer for. When im doing feature selection, i end up with a list of the top performing sets. Would it make sense to use the top 10 sets ...
0
votes
0answers
14 views

Interpreting random slope for a dataset with missing data in mixed model

I am struggling to understand the meaning of random effect for the dataset with missing data based on mixed model, I am appreciated if anyone can help. Here is an example. let us say we have 20 ...
0
votes
2answers
25 views

How you can you take the `min` of what looks like a single value calculation in SVD++?

I'm reading through this paper: http://www.cs.rochester.edu/twiki/pub/Main/HarpSeminar/Factorization_Meets_the_Neighborhood-_a_Multifaceted_Collaborative_Filtering_Model.pdf And I'm looking at the ...
0
votes
0answers
6 views

Stacking GBT using Logistic Regression

1) I build Gradient Boosted Tree Model in h2o and now i have the POJO. 2) I extracted the weight for each tree of GBT for my population 3) I used the extracted weight to train a logistic regression ...
0
votes
1answer
36 views

Model Underperforming

I am a quite new to machine learning but I have tried to implement some prediction on a data to predict if a customer would churn of not.And for this I have used many features but I am unable to ...
0
votes
2answers
39 views

Can we compare classifier scores in one-vs-all/one-vs-many?

In a system where we perform multi-class classification via a one-vs-all technique, are two scores comparable? E.g.: If I have 0.5 and 0.6 on two different classifiers, is it possible to say that the ...
1
vote
2answers
27 views

Boosted Trees classification

I'm using R's gbm() package to do a boosted classification problem, where my response variable is a binary variable taking values of 0 and 1. I have 11 predictors in my data set. After running the ...
0
votes
0answers
26 views

How to learn about adaptive neuro-fuzzy inference system (ANFIS)?

How can I learn about adaptive neuro-fuzzy inference system (ANFIS) for statistical analysis? Is there any tutorial available, or any other sources?
2
votes
3answers
67 views

Hyperparameter optimization with random search

I would like to do a random search for hyperparameter optimization. The procedure can be found in link. One possibility is to define a fine grid and take random combinations. A better approach would ...
0
votes
1answer
38 views

Are all generative Models based on Bayes?

Reading about deep learning I encounter various different kinds of hierarchical networks, many of which are generative. 1) Are all of the generative networks based on Bayes? 2) If not, how do they ...
0
votes
0answers
24 views

Neural networks that can reach state-of-the-art accuracy with two or three hours training?

Are there some neural networks that can reach state-of-the-art accuracy with two or three hours training, on dataset like CIFAR, MNIST,etc...
0
votes
1answer
25 views

Comparing and evaluating win probabilities in sports from different settings

Background I'm trying to predict the probability that the home teams wins a certain sports game, for each minute of the game. Taking these win probabilities together produces a nice visual of the ...
0
votes
1answer
21 views

Applying filters learned from convolutional neural networks

I have a neural network that I trained on 32 * 32 px size images. Can I use these filters learned from the network on larger images not used in training the network such as a 600 * 800 px image? Or ...
0
votes
0answers
6 views

finding and comparing temporal and location-specific pass patterns in a soccer game

I have data on several soccer matches, where ball passes have been recorded in terms of XY-location where the pass starts, XY location where the pass ends, the team making the pass, the player making ...
0
votes
1answer
39 views

true negative is 0% whereas true positive is 100% correctly classified

I used Naive Bayes from Spark's MlLib to train a model and test it on the data (in the form of an RDD). The results were confusing. the data and results are as follows: The problem is a binary ...