Questions tagged [unsupervised-learning]
Finding hidden (statistical) structure in unlabelled data, including clustering and feature extraction for dimensionality reduction.
641
questions
2
votes
1
answer
14
views
Image Clustering (Unsupervised learning) on unknow class(guess less than 300)
I have 30000 unlabeled images (each image has only one character), and the content of the images is very simple, basically black lines(a language but not English) and white background.
I hope to use ...
- 21
1
vote
0
answers
64
views
Chicken and egg problem in machine learning [closed]
Recently, I went through an ICLR paper SELF-LABELLING VIA SIMULTANEOUS CLUSTERING
AND REPRESENTATION LEARNING. In the paper, authors discussed simultaneously labeling the images and training a network ...
- 51
0
votes
0
answers
15
views
Any reason for choosing t-SNE over UMAP when visualizing?
According to the UMAP paper:
Our algorithm is competitive with t-SNE for visualization quality and arguably preserves more of the global structure with superior run time performance.
paper
It seems ...
- 143
2
votes
0
answers
44
views
Restricted Boltzmann Machine: W matrix visualization results after training MNIST images and Pseudo-log-likelihood
I am implementing RBM from scratch using Tensorflow and after training my RBM on the MNIST dataset for 200 epochs using Persistent CD with two steps of contrastive divergence, I learn the weights W ...
- 21
0
votes
1
answer
47
views
Do we need to split the data for Unsupervised Anomaly Detection?
I'm struggling with understanding the concept of splitting data for unsupervised anomaly/outlier detection. You can find all approaches here. I found some papers and implementations that didn't split ...
- 385
0
votes
0
answers
25
views
Maximizing a unique trace quadratic form
I am dealing with an unsupervised problem where I have ended up with the following maximization problem:
$\max_{C\in \mathbb{R}^{p\times n}}\sum_{i=0}^{m} tr(CA^ixx^\top A^{i^\top}C^\top) \\ \mathrm{...
- 43
0
votes
1
answer
33
views
In which category falls a mix of unsupervised and supvervised learning?
Here is the context of my problem:
I want to classify between to classes. However, I have at disposal only non labeled data to do the training (the test set possess all labels for evaluation purposes)....
- 9
0
votes
1
answer
19
views
Why does single linkage create loose clusters when it uses smallest distance between two points?
The definition of single linkage says: In single linkage method, the distance between two clusters is defined as the minimum distance between two data points in each cluster.
However, different ...
0
votes
1
answer
52
views
Can all neural network layers be used as either a supervised or an unsupervised model?
I am trying to understand neural networks and by reading different articles I always find conflicting information. I wanted to understand which neural networks can be used as supervised/unsupervised. ...
- 116
1
vote
0
answers
29
views
Testing if one sample can be from a specific class
Suppose I have 2 customer segments in a bank (private class and another class) defined by the institution itself. and I want to increase the amount of private clients class, bring clients from other ...
- 111
0
votes
1
answer
29
views
A method for clustering 1D signals?
I have samples from 150 different genes containing the following information:
sequence of the gene
signal strength along the length of the gene (the signal can be negative or positive).
I have ...
- 5
0
votes
0
answers
8
views
analytically determine how many clusters you need to get an explained variance of over x%
I am currently trying to cluster my data with as few clusters as possible. I have tried using K-means clustering and spectral clustering. Both work relatively well, around 85% explained variance from ...
- 1
0
votes
0
answers
11
views
Are Hidden Markov Models the right tool for signal segmentation task?
I have a particular problem, and I would like to know if using a HMM is the correct tool for it. Apologies for the poor wording of the problem, HMMs are definitely not my specialty.
I have the ...
- 101
2
votes
0
answers
40
views
Are the distances on a hierarchical clustering dendrogram in the same units as the input distance matrix?
I use Aitchison distance as the input to a hierarchical clustering dendrogram. I started labeling and interpreting the dendrogram but wasn't sure about a few aspects:
Are the vertical distances on ...
- 1,280
0
votes
0
answers
10
views
Embedding extraction -> Classifier VS Embedding learning+ Classification on-the-fly?
I have two questions:
How should we compare in general which of the following perform better?
I have a graph and would like to perform a graph classification task.
Is it better to extract graph ...
- 173
2
votes
0
answers
28
views
What is the proper way to externally validate clusters when I have only a sample of the dataset labeled, but want to cluster the entire dataset?
I have a dataset of text-based documents that I want to cluster. For a sample of this dataset (~10%) I have manually annotated labels (i.e., the ground truth). I would like to cluster this dataset to &...
- 21
0
votes
0
answers
11
views
mixture of finite regressions without a response variable
In finite mixture modelling, in particular mixture of regressions modelling, we are interested in finding latent trajectories against a response variable. But what if there is no known response ...
- 155
1
vote
0
answers
22
views
is there neural network architecture for unsupervised learning for topic modelling?
Let say i have a bunch of document and I want to analyse the topic of the corpus. The only way i know is to use unsupervised learning with gensim which use model like Latent Dirichlet allocation(LDA),...
- 11
1
vote
0
answers
27
views
What is the best approach: Labeled training data and unlabeled test data [closed]
I'm new into the data science world and I am working on improving my knowledge so here is my problem:
I want to build a binary classifier with the following constraints:
I have 2 files training.csv ...
- 11
2
votes
0
answers
73
views
PCA explained variance and model inertia
I'm trying to perform a PCA to reduce the dimensionality of my data and subsequently perform a K-Means algorithm.
I initially chose 4 Principal Components because they explain 70% of my variance. This,...
- 21
1
vote
1
answer
33
views
Why are my classification outputs sometimes saturating? [closed]
I have built a binary classifier (FWIW - using Keras) but the output values saturate at a value considerably < 1 (see plot).
Is this likely to be implementation problem? Is it because the model is ...
- 125
1
vote
0
answers
24
views
Can a Supervised Routine be Compared Against an Unsupervised?
Just a question out of curiosity. Suppose that I had generated:
(1) an unsupervised decision tree using 'interpretable clustering,' and (2) a second supervised decision tree (whether CART, or a ...
- 174
1
vote
1
answer
78
views
why is unsupervised learning said to be learning probability distribution?
In the book Deep Learning by Ian Goodfellow et al., it is mentioned that
unsupervised learning involves observing several examples
of a random vector x, and attempting to implicitly or explicitly ...
- 167
0
votes
0
answers
20
views
Correlation among categorical variable and unsupervised model with practical example
I'd like to demonstrate that men will apply for a job if their skills match at least with the 60% of the requirements in job advertisement; women only if they match 100%. I'll use R. I'll give you a ...
- 1
0
votes
0
answers
50
views
How to save a Higher accurate K-means Model on a unlabelled data based on Any Performance Evaluation Metrics?
I am experimenting on Iris dataset. I am not using the label. I want to save my model based on any Performance Metrics. According to Performance Metrics which model have higher score I am choosing ...
- 51
1
vote
0
answers
13
views
Fast Approximate Spectral Clustering in Practice?
Spectral clustering takes $O(n^3)$ time. Over the last fifteen years, a huge number of heuristics have been published on speeding up spectral clustering (e.g. Nystrom method, PCA, sample size ...
- 21
0
votes
1
answer
32
views
Can time series forecasting be done without splitting the data into train/test sets?
The data is of monthly average rainfall for a specific region for the past 13 years (156 data points). What is the best way of splitting it into train/test sets?
I thought of selecting first 12 years ...
- 65
0
votes
0
answers
29
views
Unsupervised clustering with a categorical with tens of thousands of levels
I need to perform a clustering analysis of a medical claims dataset to identify anomalous healthcare providers.
My dataset contains a variable called diagnosis code ...
- 549
0
votes
1
answer
22
views
Unsupervised Clustering with Extra Information
I have a task to cluster an almost entirely unlabelled dataset. After reading the literature on semi-supervised clustering, I have not found any algorithms that suit my very particular needs.
...
- 303
0
votes
0
answers
17
views
Finding independent variables from tabulated dataset
Given a tabulated dataset for $n$ variables, how can I find the smallest subset that consists of $m$ independent variables ($m \leq n$), so that the complete dataset can be constructed by knowing only ...
- 101
15
votes
2
answers
666
views
Finding category with maximum likelihood method
Let's say that we had an information for men and women heights.
R code:
...
- 153
1
vote
0
answers
56
views
How to calculate Maximum Adjusted Rand Index for cluster number bigger than ground truth cluster number
I am using the Adjusted Rand Index (ARI) for assessing clustering solutions against ground truth. I am consistently getting higher ARI results with higher cluster numbers (k=5) than the number of ...
- 11
1
vote
0
answers
15
views
In clustering, how does one estimate the distribution over the number of clusters?
In Murhpy's Machine Learning book pg 10, he says, when introducing clustering
(Given unlabelled data)
Let K denote the number of clusters. Our first goal is to estimate the distribution over the ...
- 1,145
1
vote
1
answer
59
views
Identity for K-Means Clustering
The property (12.18) from here states that $$\frac{1}{|C_k|} \sum_{i, i' \in C_k} \sum_{j = 1}^{p} \left(x_{ij} - x_{i'j}\right)^2 = 2 \sum_{i \in C_k} \sum_{j = 1}^{p} \left(x_{ij} - \frac{1}{|C_k|} \...
0
votes
0
answers
29
views
Good clustering method for compact clusters
Suppose I have a LARGE dataset with unknown labels (around 40000 points) where the pairwise Euclidean distance does not differ too much ( mostly 11- 13, some 6-9). Which clustering algorithm should I ...
- 205
0
votes
0
answers
18
views
Hyperparameter selection in Affinity Propagation without ground truth
My goal is to implement affinity propagation for clustering a given dataset (n=12 features), and I wish to find the optimal hyperparameter value (preference) allowing an educated guess of the number ...
- 11
1
vote
0
answers
26
views
K-Means clustering technique for monthly data
I have an Unsupervised problem where user's Credit Card payment data is given for each month for various users for one year. One of the feature in the data having "User Id". For most of the ...
0
votes
0
answers
74
views
What is the purpose of covariance-matrix in CMGOS?
I'm working on the CMGOS outlier detection algorithm in the Rapidminer tool. Clustering-based Multivariate Gaussian Outlier Score (CMGOS) is a ...
- 2,391
0
votes
0
answers
18
views
Why does Non-Negative Matrix Factorization reconstructs exactly the same matrix?
I'm trying recently to get into recommender systems and almost all tutorials I find mention collaborative filtering done with matrix factorization. I found this tutorial that describes how to build ...
0
votes
0
answers
259
views
Using Silhouette Score to evaluate different clustering algorithm
I am trying to compare different clustering algorithms on a dataset and compare the model performance. Since the dataset is quite big (56 features), I applied PCA to reduce the number of features to ...
- 1
2
votes
1
answer
22
views
Can I construct a target variable out of correlates and proxies when my training data does not have the actual target variable that I need?
I want to classify customers who at risk to churn (unsubscribe). The typical path would be to have a training set of historical data that includes observations of customers who churned so that we have ...
- 587
0
votes
1
answer
435
views
How to evaluate unsupervised Anomaly Detection using k-means
I'm trying out different anomaly detection models and would love to hear opinion on my idea from somebody experienced. My goal is to perform anomaly detection with different models and to give each ...
- 3
0
votes
1
answer
208
views
Bias-Variance tradeoff with Clustering algorithms
I'm investigating the bias-variance tradeoff in well-known machine learning algorithms. However, I'm not sure this concept applies in the case of unsupervised methods such as clustering algorithms.
Is ...
0
votes
0
answers
21
views
Metric for comparing supervised and unsupervised model
I'm searching on how to compare (validate) a supervised learning model to an unsupervised one.
Let's say I have a supervised model for fault diagnosis which can tell me how accurate it is to predict ...
- 1
0
votes
0
answers
33
views
Performance differences between variational autoencoder and t-SNE
I trained a convolutional variational autoencoder on a dataset of medical images to detect anomalies. Based on the reconstruction error, I try to distinguish between normal images and anomalies. In ...
- 113
5
votes
3
answers
2k
views
How do you learn labels with unsupervised learning?
In https://huyenchip.com/machine-learning-systems-design/design-a-machine-learning-system.html#design-a-machine-learning-system-dwGQI5R, I came across the sentence:
Similarly, you can use ...
- 877
0
votes
0
answers
23
views
Unsupervised Clustering of large multi-dimentional data
Hello I am a machine learning newbie. I need some help with unsupervised clustering of high dimentional data. I have data with over 15 dimensions with around 50 - 80 thousand rows. The data looks ...
0
votes
1
answer
16
views
When using "silhouette coefficient" to evaluate an unsupervised model, do we need a labelled dataset?
How does "silhouette coefficient" can find the optimal number of clusters when the dataset is not labelled? Does it need a labelled dataset or is it pure statistics?
I mean, when doing ...
- 1,019
0
votes
0
answers
22
views
Clustering machine learning algorithms
I am looking for a clustering algorithm that can separate my data into clusters. The data has three different variables and there are approx 500k samples. I need some control over the number of ...
- 21
1
vote
0
answers
13
views
select the final clustering based on agreement of trials between two groups
There are well-recognized cross-validation based methods to select the number of clusters (e.g. in this answer) . However, suppose I know the number of clusters beforehand, can I select the final ...
- 45