Questions tagged [online]

Online algorithms refer to computations that are performed iteratively, with data arriving during the computation. For questions focusing on the Internet, please use the "internet" tag.

Filter by
Sorted by
Tagged with
0
votes
0answers
13 views

Batch computation vs online computation

In short, could you explain in which situations online computation is better than batch computation? (I am currently reading a paper (https://arxiv.org/abs/1003.0120) about offline policy evaluation ...
1
vote
0answers
19 views

Online Optimization - Regret in Absolute Error

In the online convex optimization literature static regret is defined as $\sum_{t=1}^{T}\left(f_t\left(x_t\right)-f_t\left(x^*\right)\right)$ where $x^*=\arg min_{x\in\mathcal{X}}\sum_{t=1}^{T}f_t\...
-2
votes
1answer
68 views

General process of training a Neural Network [closed]

I have a very broad question about the general procedure of training a NN. I am not too concerned about the precise algorithms in this question at the moment. But there is one thing bothering me. ...
0
votes
1answer
34 views

What is the cumulants of a whole data in terms of the cumulants of its parts?

I have around 8 billion data points, and I need to calculate the distribution and the cumulants of this distribution. However, due to technical restrictions, and time constraints, I can only ...
2
votes
1answer
77 views

Can you use stochastic gradient descent with a multinomial likelihood?

I have a multinomial likelihood of the form: $$P(\underline n|\underline x) = N!\prod_{i=1}^M \frac{f_i(\underline x)^{n_i}}{n_i!}$$ where $\underline x$ is a vector of parameters, $f_i(\underline x)...
0
votes
0answers
11 views

Does Follow-The-Leader incur log(T) regret on quadratic loss functions?

I encountered this statement in multiple lecture slides I found through duckduckgo, but I found no proof, and it doesn't seem trivial for the general case. Can anyone verify the $log(T)$ regret bound ...
2
votes
1answer
44 views

Online update of Pearson coefficient

Suppose I have an online stream of data points $x_i,y_i$, where $i=1,2,\dots$. I want to compute the Pearson correlation coefficient between the vectors $\vec x$ and $\vec y$. But here is the catch. ...
1
vote
1answer
49 views

Resources for Non-Bayesian Online Change-point Detection

I am interested in learning about the fundamentals of on-line change point detection. I am specifically not interested in the Bayesian methods. The only solid resource / review / survey I could find ...
0
votes
0answers
8 views

Probability of Incurring Maximum Loss

In online classification one can use mistake bound learning, where one assumes that all $y$ are generated by some target mapping $h^*: \mathcal{X} \rightarrow \mathcal{Y},\,\, h^* \in \mathcal{H}$. ...
1
vote
1answer
31 views

Testing methodology

I recently built a simple feed forward NN to predict daily demand (48 output neutrons, representing half hours) based upon 32 input features. I tested the performance by firstly doing 10 fold cross ...
2
votes
1answer
30 views

Is there a goodness of fit metric that can be computed online with $O(1)$ memory?

Say I have two random streams of two dimensional data. I want to measure how closely their underlying PDF's match. My current method is to estimate the PDF's by accumulating the samples online in a ...
1
vote
0answers
54 views

Online learning in Bayesian Linear Regression [closed]

Bishop's machine learning textbook describes ways to conduct frequentist Linear Regression, and also examines sequential updating (online learning) in this context. The textbook also covers Bayesian ...
0
votes
0answers
16 views

Dynamic Calculation of Variance after removing an observation

I know that there are formulas for calculating variance in dynamic fashions. I have also seen formulas for calculating rolling variances assuming you know what number is leaving and entering the ...
0
votes
0answers
34 views

Is there a name for this? Updating class probabilities, online

An object is of an unknown class $y$. We receive a stream of measurements $x_1, x_2, ...$ of the object. Every time we receive a new measurement, we want to update our estimate of the class label $y$ (...
15
votes
2answers
585 views

Recursively updating the MLE as new observations stream in

General Question Say we have iid data $x_1$, $x_2$, ... $\sim f(x\,|\,\boldsymbol{\theta})$ streaming in. We want to recursively compute the maximum likelihood estimate of $\boldsymbol{\theta}$. That ...
3
votes
1answer
51 views

what is the diffrences between online and one pass learning?

as long as I know, online learning takes actions at each time step (for each data), and one-pass algorithm just can see each data once. I already read Wikipedia: about streaming algorithms. These ...
0
votes
1answer
50 views

How to use/treat a hidden layer as the new target to predict in a neural network?

Let's suppose I have a neural network with one hidden layer. During training, for a given pair of (input, target), I want to perform several iterations, such that the first iteration would be trying ...
1
vote
0answers
34 views

Clarification regarding proof of convergence of online EM

Online EM algorithm was proposed by Olivier Cappé in Link to paper. They assume that complete data likelihood $f(x ; \theta)$ belongs to exponential family i.e. $f(x;\theta) = h(x) \exp \left\lbrace ...
0
votes
0answers
19 views

measuring constant value

I'm taking multiple measurements of the same constant value. The measurements are noisy (lots of outliers) and I can take lots of them, more than I can store. How would I approximate the true value? ...
2
votes
1answer
82 views

Tuning distance threshold in online clustering

In online clustering there are approaches where a threshold $r$ on the distance to the nearest cluster is used to determine whether a new data point should be associated to an existing cluster or ...
1
vote
0answers
39 views

What is the relationship between Online Learning and Statistical Learning?

Online Learning also known as Online Convex Opimization has famous algorithms like Follow-the-Leader and Online Gradient Descent (See OCO Book) Now stoastic programming has algorithms like Sample ...
1
vote
2answers
63 views

on-line regression with 1 output [closed]

I have 12 input variables from sensor (IMU) to predict 1 output (Speed of a boat) variable. Is it possible to use regression (or something else?) in this case where it is a continuous data stream from ...
4
votes
2answers
110 views

Practical realities of updating a trained model with new data

In my day to day work, I train models on data using R packages that have no extension for Bayesian priors. I will generally have a large dataset to start off with, and add new data as needed. Any ...
0
votes
0answers
19 views

Are there specific Machine Learning Algorithms that are more indicated for Real Time Analytics?

As the title suggests, I am wondering if there are specific ML algorithms that are more suitable for real time learning. In my case, I am working on deploying a stacking algorithm on Spark Streaming ...
1
vote
1answer
68 views

Regret in online learning

In online learning/online convex optimization, it's often the case that you compare your algorithm against the best action in hindsight (i.e., from https://people.cs.umass.edu/~akshay/courses/cs690m/...
1
vote
0answers
26 views

Transforming an echo state network to online training

I have recently become interested in echo state networks, I've been using the simple introduction to ESNs here https://mantas.info/wp/wp-content/uploads/simple_esn/minimalESN.py by Dr. Mantas ...
1
vote
0answers
26 views

What is the effect of the use of online-learning algorithms on non streaming data?

I am wondering what the effects of using a passive-aggressive classifier instead of something like a SVM classifier on a non-streaming data would be. In other words, what are the general assumptions ...
0
votes
0answers
16 views

Recursive least square with incomplete data

If we have a system $\mathbf{Y}_t = \mathbf{A}_0 + \mathbf{v}_t$, in which $\mathbf{v}_t$ is the noise. Using the first $t$ samples, we can estimate $\mathbf{A}_0$ by the arithmetic (sample) mean, ...
2
votes
2answers
906 views

How to frequently update classification model with new training data?

How do I incorporate a new stream of data into my classification model? Do I have to retrain the model from the beginning every time I want to incorporate new data, or can I update the existing model ...
1
vote
0answers
26 views

Any previous work on online coordinate descent, where a new coordinate appears at each iteration?

I'd like to analyze the behavior of coordinate descent algorithms, where as a twist, at each iteration, a new variable appears. For example, if $T$ is my total number of iterations, then at iteration $...
1
vote
2answers
106 views

What test to use to find the probability of highest value?

If I have a vector of around 40 values each with a normally distributed error, is there an easy way to figure out the probability of each element being the element with maximal true value? For context,...
1
vote
0answers
27 views

Online estimation of drifting discrete probability

I recently come across (in a practical setting) to the following problem. Suppose I receive items from a finite set ,one at a time . At each moment one item is drawn independently from an unknown ...
1
vote
0answers
29 views

Performance of Hierarchical Temporal Memory on unsupervised online anomaly detection problems

I'm looking for an algorithm to detect anomalies in streaming data (server metrics). The detection needs to be near-real time and unsupervised (labeled data will never be available, unfortunately, and ...
1
vote
0answers
42 views

Adding new center in an RBF network without memorizing previous training examples

Suppose we train an RBF by minimizing the LSE on a couple of training points and we are doing it incrementally in an online fashion. So basically we update the QR factorization using e.g. Givens ...
1
vote
1answer
41 views

Algorithm for selecting largest possible value, when observing online sequence of unknown distribution?

I have been trying to devise an algorithm for a problem that's been bugging me for a while. For some weird reason I haven't been able to find any mention of this problem in the literature, so far. I ...
1
vote
2answers
40 views

Effects of exclusion on averages

Given a set of data points, if a single point which is above/below the average is excluded, can it be said that the new average will surely decrease/increase? What if the average is the median instead ...
1
vote
0answers
56 views

Extended kalman filter vs online passive-aggressive

I was wondering, what are the advantages and disadvantages of extended Kalman filter and online passive-aggressive algorithm when we use them to train our networks. I have RBF neural network and I'm ...
3
votes
1answer
46 views

Are ensemble learning methods for data streams restricted to online or batch learning?

Recently I'm working on some online learning algorithm (using RBF neural network ) for classification. As I read papers in this area I found there is an issue in online-learning called concept drift ...
0
votes
1answer
62 views

Real-Time signal processing with RandomForestClassifier in sklearn always predicts one class

I am trying to perform real-time decision making on data from a radar sensor and trying to detect occupancy. I generated data using the same sensor annotated it manually as vacant or occupied. I ...
1
vote
1answer
63 views

What data format should I use to learn the nonlinear output behavior of my guitar distortion pedal using a neural networkl?

My Problem I've built a very simple transistor guitar pedal. it has 1 mono input, 1 mono output. Now, all I have ever done in the past with ANN's is offline learning with labelled data and some work ...
4
votes
1answer
50 views

Gradual clustering in deterministic manner

We have 128-dimensional vectors representing people's identities where the euclidean metric defines the similarity between them. Ours solution requires them to be clustered and then annotated (assign ...
1
vote
0answers
120 views

What are online learning and offline learning in the context of reinforcement learning? [duplicate]

In this question, many users have discussed online and offline learning in machine learning. But, in the context of reinforcement learning, what are exactly online and offline learning?
1
vote
0answers
295 views

Updating of OLS regression estimates when one data point is changed

How do I estimate the new slope and intercept if I'm given a 'updated' data instance? For example I have a regression equation y = 1x + 0.5 and this is learned with a data set of 10 data instances ...
4
votes
3answers
1k views

Online algorithm for the mean square error

Given a dataset $\{(x_1, y_1), (x_2, y_2), \dots\}$, we can compute incrementally (or "online") the linear regression for those points. In other words, given a new point $(x_i, y_i)$, we can ...
1
vote
0answers
76 views

Online deseasonalization of time series data

Are there any existing methodologies of deseasonalizing time series data online, in order to avoid lookahead bias? It seems that if you don't deaseasonalize time series data online, you would not be ...
1
vote
2answers
31 views

Learning from Mistakes in a classification task

Let's say I have a text classifier that I have trained on some training dataset. Now when I run this trained classifier on some test dataset, I identify the cases where it went wrong (assuming I have ...
0
votes
1answer
49 views

Calculating mean and SD from a stream of numbers

I'm sure this has been asked, but my stats knowledge is not good enough to know the terms to looks for, so apologies if this is really basic. I'm collecting a sample of numbers for which I want to ...
2
votes
1answer
102 views

Matrix factorization for expanding matrix

In the paper Matrix Factorization Techniques for Recommender Systems Koren, Bell and Volinsky describe how the matrix $R_{n \times k}$ (users $\times$ movie ratings) can be decomposed to $P_{n \times ...
1
vote
1answer
239 views

Algorithms for real-time classification of segments in noisy time-series data

I’m trying to detect various features of a toy train track while driving on it: The primary input is data from an optical sensor. The following image shows the recorded signal when driving over the ...
1
vote
1answer
73 views

What are alternatives to regret minimization in online learning settings?

Regret is a common criteria to optimize in online learning. I'm wondering if anyone knows of other alternative criteria to optimize that have been proposed or explored in an online learning problem. ...