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

learn more… | top users | synonyms (1)

0
votes
0answers
13 views

Fractional Model with Instrumental Variable in panel data

I faced econometric problem in my estimation strategy. My explanatory variable is a proportion (i.e. the percentage of internal finance in working and fixed capital finance) and at the same time, it ...
0
votes
0answers
24 views

Combining collaborative filtering and market basket analysis for effective recommendation

I am working on a store data set and I am trying to figure out ways to come up with more effective recommendations based on Market basket analysis and Collaborative filtering methods. Is it possible ...
2
votes
0answers
82 views

Relu vs Sigmoid vs Softmax as hidden layer neurons

I was playing with a simple Neural Network with only one hidden layer, by Tensorflow, and then I tried different activations for the hidden layer: Relu Sigmoid Softmax (well, usually softmax is used ...
0
votes
1answer
72 views

Partitioning training data for dimension reduction and classification

Let's say I want to test the performance of my dimension reduction + classification pipeline. To do this, I will use k-fold cross validation. I know that performing dimension reduction on the complete ...
0
votes
0answers
22 views

Does the field of dynamical systems have applications in machine learning?

Does the field of dynamical systems have any applications in machine learning? I have taken a short module on dynamical systems in the past but if there is a possibility it has worthwhile applications ...
0
votes
0answers
37 views

LogisticRegressionCV or LogisticRegression

I'm new worker with scikit-learn and would like to know when to use LogisticRegressionCV and when to use LogisticRegression. Thanks.
1
vote
1answer
53 views

loss function for data with noisy labels?

I'm trying to train a neural network for classification, but the labels I have are rather noisy (around 30% of the labels are wrong). The cross-entropy loss indeed works, but I was wondering are ...
1
vote
0answers
21 views

Evaluation of a ternary classifier

Are there standard evaluation procedures for non-binary classifiers? In my case I have "nested" classes, being absence and presence of an effect the first and usual binary categorization, but ...
0
votes
0answers
7 views

Many class, one train sample per class classification

I have a classification task involving one train sample per class with around 300 classes.Furthermore each observasion has about 200 features. Can anyone suggest an approach that might work better ...
0
votes
0answers
79 views

Recommended Data Augmentation Techniques for Deep Convolutional Neural Networks

Introduction&Background: My experience with deep learning research has shown me that data augmentation is one of the most important techniques one can use to improve performance (unfortunately?)....
1
vote
0answers
31 views

How to design a fitness function for an evolving neural network?

I'm working on making my own neural network using the NEAT algorithm. I have programmed the algorithm from scratch because I can't seem to get any of the libraries online working, but I'm 90% sure the ...
1
vote
1answer
44 views

Why take the gradient of the moments (mean and variance) when using Batch Normalization in a Neural Network?

When doing Batch Normalization (BN) it makes sense to me to treat the BN transform as a layer that we need to do back propagatiod and thus have derivatives to update its parameters (for each layer) ...
2
votes
0answers
26 views

Why are there three parameters that can change during batch normalization according to MatConvNet implementation?

I was going through the MNIST example that comes with MatConvNet were they attempt to demonstrate how to use the API of their Batch Normalization. However if you take a look at the training file there ...
1
vote
2answers
57 views

What is wrong with “PROC fishing” syndrome?

RE: https://www.quora.com/What-do-statisticians-e-g-Stats-PhDs-think-of-data-scientists-in-industry-without-stats-backgrounds There are several comments made regarding "PROC FISH syndrome", whereby ...
2
votes
0answers
41 views

What does MatConvNet do with Batch Normalization during testing and inference?

I was reading the documentation for evaluating a simple CNN and it said: In test mode, dropout and batch-normalization are bypassed. Note that, when a network is deployed, it may be preferable ...
1
vote
0answers
12 views

Applying DBSCAN to a huge GIS dataset with a Haversine distance metric.

I have a training set (2GB) that contains GIS trajectory data for multiple taxi rides. I want to cluster the final destinations based on their spatial density and have therefore been trying to use the ...
0
votes
0answers
17 views

Clustering on two feature spaces

I want to do a clustering task with my data to find some interesting patterns. I have two groups/vectors of features V1 and V2 which are fundamentally different and are supposed to measure different ...
1
vote
2answers
39 views

Analysising the difference between k-means and spectral clustering algorithm

My target is to cluster the spatial area based on the location(X,Y) and pollutant concentration(Z). So there would be three different attributes along the spatial area(n_sample = grid point) I have ...
0
votes
2answers
68 views

Seeking a Recommendation on Machine Learning Books for Biological Research

I am an undergraduate student studying mathematics and microbiology. I recently got a research project to study the evolution of viruses from the computational perspective, particularly from machine ...
3
votes
2answers
68 views

How does Naive Bayes work with continuous variables?

To my (very basic) understanding, Naive Bayes estimates probabilities based on the class frequencies of each feature in the training data. But how does it calculate the frequency of continuous ...
1
vote
0answers
51 views

Modeling the importance of data points in Logistic Regression

Given N data points where each entry of a point represents a value of a feature, and we need to use this data to model a binary Logistic regression model. e.g. lets say the data points represents ...
0
votes
1answer
55 views

Streaming data input in artificial neural network

Suppose we have continuous stream of data which length we cannot predict and discretize. Is there a type of neural network that can hold this stream and makes output based on the information stored in ...
2
votes
0answers
18 views

Is it possible for the forward pass computed by a wrapper to not agree with a hard coded using MatConvNet?

I was implementing a super duper simple network and I wasn't able to have the forward pass match for both the wrapper and the hardcoded version of the network: ...
2
votes
0answers
35 views

How can I compute statistical significance with cross-validation? [duplicate]

I've done nested cross-validation with 10 outer runs and 10 inner runs to evaluate different classifiers. In the end I average the accuracies of the 10 outer cross-validation runs to get a final ...
1
vote
0answers
43 views

Why do boosting overfit on the data with uniform noise?

i read about it but i didn't get the idea, and actually i didn't find many pages that talk about uniform noise with boosting, is it rare to happen or what? another question: i read in some pages that ...
0
votes
0answers
18 views

Is there a one-class hinge loss function?

Is there a [hinge] loss function that is suitable for one-class classifiers? i.e., anomaly detection. Thanks.
3
votes
1answer
58 views

What happens if I flip targets and predictions in cross-entropy?

When we compute the cross-entropy within the machine learning context, we use the following formula: $$ CE(t, p) = -\sum_{i=1}^{N} t_i \ \log(p_i) $$ Where $t$ is the target class probability, and $...
1
vote
0answers
28 views

Is it possible to use Caffe to classify text?

I'm trying to build a sentence classifier using Caffe, for example the following dataset: 1 -> 'Fell energy at my home' 2 -> 'I liked the city's new buses' 3 -> 'It has a hole in the road in front ...
0
votes
1answer
19 views

Explanation of different testtype and teststats in ctree in party package of R

I was looking into the ctree function in the party package for R. In the ctree_control ...
1
vote
0answers
10 views

Is it needed to regularize in case you know your data is generated by a model of your model class?

Assume we have a dataset $X_{full}$ with labels $y_{full}$. We train a kernel ridge regression model on this data with the Gaussian kernel. This model is used to generate predictions on the whole ...
3
votes
1answer
119 views

ideas on machine-learning algorithms to classify products

I have a list of products, including variables such as the product name (as it appears on the receipt) and the merchant where the product was bought. I have a good deal of them manually classified ...
0
votes
0answers
26 views

Methods for time-series pattern weightings in machine learning

I am currently developing an anomaly detection algorithm to test for anomalies in system usage. This is obviously very seasonal in nature (for example, during the Super Bowl, for example, there will ...
0
votes
1answer
42 views

Improving a boosted regression model or change?

I am looking at a data set that contains multiple predictors and a continuous response. Using dismo along with gbm I built (a terrible one?) model. Using the package sROC, I got an AUC or 0.48 - so my ...
1
vote
0answers
28 views

How to calculate the distance in KNN for mixed data types?

when the data is from different types (numerical and categorical) of course euclidean distance alone or hamming distance alone can't help. so i have 2 approaches: standardize all the data with ...
1
vote
2answers
52 views

Does repeated k-fold cross-validation give the same answers each time?

I have $n$ instances in my data and I will do 5-fold cross validation on it (like in the picture): But when I read about "repeated cross-validation" I think that it will give me exactly the same ...
0
votes
0answers
7 views

Which will be better If I have 2 years data (training and testing) with a condition

Condition: I will always asks the model to predict the behavior of last 1 month data i.e I want the result on last 1 month of data. I have 2 years of data of my app, and I have to train the model and ...
1
vote
1answer
38 views

Why does k-fold cross validations help obtain stable clustering results?

Why applying k-fold cross validation helps obtaining stable clustering results in unsupervised learning? How is this done? Thanks
1
vote
0answers
29 views

HMM for sequence classification in R

I have a dataset which includes sequence of DNA nucleotides (A,C,G,T) and each sequence has a gene index that is binary. i.e I'm trying to classify unknown sequence by using a hidden markov model ...
0
votes
1answer
51 views

How does gamma in SVM RBF kernel influence the accuracy?

I am working on a classification program using SVM RBF kernel. To find the best parameters C and gamma, I used grid search, and got the image below. What confuses me is that when gamma varies from 0.3 ...
0
votes
0answers
87 views

XGBoost - Can we find a “better” objective function than RMSE for regression?

If we think back to linear models for a moment, we have Ordinary Least Squares (OLS) versus Generalized Linear Models (GLM). Without going too in-depth, it can be said that GLMs "improve" upon OLS by ...
2
votes
2answers
66 views

Good examples/books/resources to learn about applied machine learning (not just ML itself)

I've taken an ML course previously, but now that I am working with ML related projects at my job, I am struggling quite a bit to actually apply it. I'm sure the stuff I'm doing has been researched/...
0
votes
0answers
13 views

Combining probabilities with different amount of evidence in sequence learning

Let us think of a simple case of sequence prediction. Based on 20 observed items, b was observed 16 times, c 3 times, and d only 1 time. The sequence is as follows: ...
0
votes
0answers
27 views

Probabilities of classes using h2o.predict

How does h2o.predict calculate the probabilities of different classes ? In the randomForest package of R, the probabilities are calculated based on the number of votes. For example probability of ...
0
votes
0answers
24 views

Implementing Recommender System Ranking Algorithm in Python

I am trying to do collaborative filtering for implicit feedback datasets by following the seminal paper: http://yifanhu.net/PUB/cf.pdf The section on ranking says: I have a matrix of 50K X 9K ...
0
votes
0answers
41 views

How to create a very simple 1D example Neural Network using MatConvNet [on hold]

I was learning MatConvNet and I was trying to create a super extremely simple 1D example. I wanted to implement the following NN: $$ f(x; \theta) = \sum^{K}_{i=1} c_i ( x v_i + b_i )_+ = \sum^{K}_{i=...
0
votes
0answers
16 views

Are scalability, parallelization and online learning the same?

Someone told me before that all of them are the same in machine learning algorithms. Is that right or not? Can an algorithm be scalable but can't be parallelized? Can it be parallelized but can't be ...
0
votes
2answers
76 views

Is there a 3D neural network and how to code it in R?

I have coded some NN using neuralnet and caret packages. But these neural networks are of 2-Dimension (I dont know even this term exists) But I was curious whether "3D neural network" exists. While ...
1
vote
0answers
39 views

Isn't stacking models a direct approach to overfitting?

With help by the discussions here I successfully trained various models for classification. As an example say I trained a stochastic gradient boosted model (gbm) and an extreme gradient boosted tree ...
3
votes
1answer
73 views

Can deep neural network approximate multiplication function without normalization?

Let say we want to do regression for simple f = x * y using standart deep neural network. I remember that there are reseraches that tells that NN with one hiden ...
0
votes
0answers
26 views

An interesting two model comparison metric

My colleagues at work use an interesting metric to compare two predictive models that I have never seen before. It can be used for both classification and regression. But we consider classification ...