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

learn more… | top users | synonyms (1)

0
votes
0answers
3 views

different methods of calculating feature coefficents

I was reading up on logistic regression, and if I'm not mistaken, the algorithm for discovering the most likely coefficients is just the Perceptron algorithm. First off is this correct? If so, is this ...
0
votes
0answers
21 views

Classifier suggestion(27 dimensions, 9 classes)

The restriction of my classification problem is: 27 dimensions, 9 classes, 50.000 entries in the training set, 150.000 in test set. I need a machine learning classifier(open source code) that fits on ...
4
votes
1answer
39 views

LDA vs word2vec

I am trying to understand what is similarity between Latent Dirichlet Allocation and word2vec for calculating word similarity. As I understand, LDA maps words to a vector of probabilities of latent ...
1
vote
3answers
29 views

When does the EM for Gaussian mixture model has one of the Gaussian diminish to exactly one point and have zero variance?

I had implemented the EM algorithm for mixture models as follows: For the E-step I compute the soft-counts of assigning each point $x^{(t)} \in Data_n$ to an individual cluster $j \in \{1, ..., K \}$ ...
1
vote
2answers
44 views

Word2Vec : Interpretation of Subtraction or addition of vectors

I am curious, what does subtracting vectors, as in [man – woman] do in regards to Google's word2vec calculation of analogy ? Is this a measure of how different the two vectors are? So is man – woman ...
0
votes
0answers
23 views

Models to predict response rate of market campaign?

I have been trying logistic regression to fit the data and get an estimation of the response rate, but the power of the model is quite limited. The area under the ROC curve is always around 0.6. I ...
1
vote
0answers
17 views

Best Random Forest model converging to bagging: What does it mean? (R)

I am performing a grid search to tune the Random Forest parameters m and nodesize. I have 79 variables, and the best model, in terms of OOB error, is a model with 76 variables (OOB error = 0.137). So, ...
0
votes
1answer
23 views

How much prediction accuracy of SVM (or other ML models) depend on the way features are encoded?

Suppose that for a given ML problem, we have a feature which car the person possesses. We can encode this information in one of the following ways: Assign an id to each of the car. Make a column ...
2
votes
1answer
27 views

Zero-inflated gamma - how to write down the cdf?

My goal is building a predictive model to give probabilistic forecasts. My response variable has lots of zeros but otherwise looks close to a gamma. I fit the whole dataset using some classification ...
0
votes
0answers
20 views

What to expect when training deep neural networks with increasing capacity?

I have a question with regards to the order in which to go about when trying different deep neural network architectures for a task. Suppose I trained a model with $|P_1|$ parameters, and noted down ...
0
votes
1answer
22 views

Validation of Kernel Regression

I have a data set which I divided in two part(development and validation). I am using kernel regression for build a relationship between independent and dependent variables in development period data. ...
1
vote
0answers
30 views

How do you do EM algorithm for a factored model for a recommender system?

Let $X$ be a $n \times d$ matrix with users as rows and movies as columns. Each user is a single row $x^{(u)} \in \mathbb{R}^d$ (i.e. for user u there are at most d ratings for the d movies). Also ...
0
votes
0answers
69 views

Random Forest accuracy 0.98, is it too much?

I am using about 256 predictors and target is sales. I am using a software called Alteryx which is R based. I have tried to run Random Forest, Spline model and Neural nets on same data. I used ...
0
votes
1answer
59 views

What will Multi-variate Gaussian Distribution become when the quadratic form in the exponent is constant

I'm reading the chapter 2 of PRML(Pattern Recognition and Machine Learning) and I don't understand the following statement in the first paragraph in page 80: The Gaussian distribution will be ...
0
votes
0answers
9 views

basic implementation of Non-parametric Bayesian model in python

I am having a problem in understanding infinite Bayesian model with its implementation. I have tried looking scikit-learn package of python, DPGMM. I dont know why there is an argument for defining ...
0
votes
0answers
11 views

Robot localization and position prediction

The problem I am working on is related to robot localization and position prediction in the future. Given a simple video of a robot bouncing around in a wooden box and a mapping of the coordinates ...
1
vote
1answer
25 views

Is there any package in R that's commonly used for semi-supervised learning?

Is there any package in R that's commonly used for semi-supervised learning ? I have a dataset where I manually labeled 100 data points so I'd like to use semi-supervise learning for the rest of the ...
0
votes
0answers
28 views

How to decide about Regression Analysis or Time Series Analysis

How to decide about Regression Analysis or Time Series Analysis. ...
1
vote
0answers
19 views

What machine learning algorithm should I choose to fill in blanks from context?

I have a project where I need to be able to fill in a missing word given a few words of context. In other words, suppose I have a sentence: I went ____ the store. I want to be able to deduce ...
2
votes
1answer
443 views

Why is n-gram used in text language identification instead of words?

In two popular language identification libraries, Compact Language Detector 2 for C++ and language detector for java, both of them used (character based) n-grams to extract text features. Why is a ...
1
vote
0answers
16 views

“interact.gbm” in package gbm vs. “gbm.interactions” in package dismo

The reference manual for the gbm package states the interact.gbm function computes Friedman's H-statistic to assess the strength ...
0
votes
0answers
49 views

How are tensors used in neural networks?

I'm new to learning neural networks and am trying to understand the role of tensors in them. I am trying to use some neural network libraries, but they are asking me for the dimensions. Could any one ...
0
votes
0answers
16 views

What is the VC Dimension of a Naive Bayes Classifier?

How do you calculate the VC dimension of a Naive Bayes classifier with say K features?
0
votes
0answers
8 views

References for learning text stemming

I am trying to learn and experiment with text stemming. My ultimate goal is knowledge extraction from scientific text and corpus with emphasis on contextually multiplicity. But text stemming and ...
0
votes
0answers
50 views

Time Series Analysis

I have attended a lecture about introduction to machine learning at my university (SVM, regression, kernels etc.). Now I'm planning to do a project where machine learning knowledge is needed. In fact ...
6
votes
1answer
64 views

“Export” machine learning model from R

I can build and implement classic ML models on traditional training/test sets in R, but what if a partner wants to get this model in order to implement his own (any kind of) system? Saving and sending ...
0
votes
0answers
7 views

How can one incorporate prior knowledge about the distribution to which a feature conforms into a machine learning model to improve its performance?

For example, if I were building an SVM to predict whether or not a person is a professional basketball player, and I knew that height (one of the features available) were normally distributed, how ...
0
votes
0answers
17 views

PDE VS machine learning when solving complex systems?

I am wondering how PDE can be used in machine learning theory. I have got idea from this post also this question Based on what I learn from machine learning discriminative and generative models, I ...
0
votes
3answers
50 views

How To Determine The Number Of Dimensions To A Machine Learning Problem

I have a bit to learn about machine learning, so please pardon me if I am asking the wrong type of question. I have read some about neural networks and SVMs, so I'm not completely in the dark. I am ...
1
vote
0answers
32 views

What are the computer related prerequisite to do cool stuff with data? [closed]

I am a mathematician, who has recently gotten interested in statistics and machine learning, and feel that the biggest gap I have to fill is the technological one. What are the different ways that ...
0
votes
1answer
43 views

Can I use ReLU in autoencoder as activation function?

When we discuss about autenocder in neural network, most people will use sigmoid as the activation function. Can we use ReLU instead? (Since ReLU has no limit on the upper bound, basically meaning ...
1
vote
0answers
13 views

Relation between chi-squared statistic scores and classification accuracy

I am evaluating the utility of two distinct sets of features for solving a given supervised classification problem with two classes. I am using the chi-squared statistic as a feature selection ...
0
votes
0answers
9 views

Finding the features that have meaning in subset of data

I have a set of $N$ points $x_i=(x_i^1, x_i^2,...,x_i^{m+k})$ in $m+k$-dimensional space ($m$ continuous dimensions and $k$ discrete). Also I have a subset of these points that are marked as "bad". ...
1
vote
1answer
46 views

Maximum Likelihood Estimate (MLE) equivalent to finding $\hat y$ in linear regression with i.i.d. Gaussian noise distribution

In an assignment I need to show that for linear regression, with the noise i.i.d. Gaussian distributed $\epsilon_i \sim N(0,\sigma^2)$, that finding the Maximum Likelihood Estimate (MLE) is equivalent ...
4
votes
0answers
72 views

Are all models useless? Is any exact model possible — or useful?

This question has been festering in my mind for over a month. The February 2015 issue of Amstat News contains an article by Berkeley Professor Mark van der Laan that scolds people for using inexact ...
1
vote
1answer
20 views

Classification on variable-length time series

I have a series of transactions like the following: [0, 2, 2, 3, 1, 0, 0, 0, 1] [1, 0, 0] [3, 3, 1, 1] I would like to classify each transaction as being part of one of two categories: class A or ...
0
votes
1answer
25 views

Valid result when adding two kernels with negative coefficient?

If $k_1$ and $k_2$ be a kernel in $ \mathbb{R}^n \times \mathbb{R}^n $. we know $k(x,z)=ak_1(x,z) + bk_2(x,z)$ (kernel addition) is still a valid kernel if $\: a,b \geq 0\,$ ($a,b$ is real numbers, ...
0
votes
0answers
30 views

Cross-Validation gives different result on the same data

I have done Cross-Validation by crossval function in matlab on my data, but when I run the Cross-Validation many times, it give me a different results, so is that ...
4
votes
1answer
103 views

Should you tune 'ntree' in the Random Forest algorithm?

In the original paper, I was under the impression that the RF couldn't really overfit. However, in practice I'm seeing that increasing 'ntree' sometimes increases test set error. Is this due to ...
0
votes
0answers
25 views

Clustering on this reinforcement learning approach?

I am trying to create an agent that selects an action depending on a state that gives back maximum reward. To keep things simple I will keep it to two actions and 24 different states. The states is ...
8
votes
1answer
143 views

two margin comparison and one conclusion?

I read following notes, and couldn't get it. any idea or hint would highly appreciated. a SVM classifier using a second order polynomial kernel. The first polynomial kernel maps each input data x to ...
2
votes
1answer
23 views

variance in test accuracy will increase as we increase the number of test examples؟

I see this statement on 1 that say a True statement on Machine Learning Context. The variance in test accuracy will increase as we increase the number of test examples. my challenge is why ...
0
votes
1answer
15 views

1) State 2) Action and then 3) Reward diagram: Which ML approach to use?

It is looks like a reinforcement learning diagram however it's slightly different. I'll explain the numbers. 1) The environment first gives the agent a state 2) The agent does it's magic and then ...
2
votes
0answers
41 views

Supervised learning, unsupervised learning and reinforcement learning: Workflow basics

Supervised learning 1) A human builds a classifier based on input and output data 2) That classifier is trained with a training set of data 3) That classifier is tested with a test set of data 4) ...
0
votes
0answers
17 views

Which machine learning(recurrent/reinforcement learning) method/algorthim would suite this scenario

This application has it's roots in public transport, users opening the application and looking at the departure times of buses for specific stops (page 1) or planning a journey from location A to B ...
11
votes
1answer
233 views

calculation threshold for minimum risk classifier?

Suppose Two Class $C_1$ and $C_2$ has an attribute $x$ and has distribution $ \cal{N} (0, 0.5)$ and $ \cal{N} (1, 0.5)$. if we have equal prior $P(C_1)=P(C_2)=0.5$ for following cost matrix: $L= ...
0
votes
0answers
15 views

linear discriminative analysis for regression

LDA computes a projection matrix to maximize class conditional probability. Similar to this, is there any exisiting method or library for jointly learning latent space and minimizing the regression ...
0
votes
0answers
10 views

EM to Variational EM in LDA

Why exactly, when learning hidden variables distribution in LDA(Latent Dirichlet Allocation), one cannot use to the EM (Expectation Maximization) algorithm and have to resort to a variationnal EM ...
3
votes
1answer
82 views

some inference about k-NN algorithms for better understanding? [closed]

I ran into some facts make me confusing. for k-NN classifier: I) why classification accuracy is not better with large values of k. II) the decision boundary is not smoother with smaller ...
2
votes
1answer
15 views

What is the difference between linear perceptron regression and LS linear regression?

Recently, a project I'm involved in made use of a linear perceptron for multiple (21 predictor) regression. It used stochastic GD. How is this different from OLS linear regression?