clustering a SUPER large data set [closed] - database

Closed. This question needs to be more focused. It is not currently accepting answers.
Want to improve this question? Update the question so it focuses on one problem only by editing this post.
Closed 9 years ago.
Improve this question
I am working on a project as a part of my class curriculum . Its a project for Advanced Database Management Systems and it goes like this.
1)Download large number of images (1000,000) --> Done
2)Cluster them according to their visual Similarity
a)Find histogram of each image --> Done
b)Now group (cluster) images according to their visual similarity.
Now, I am having a problem with part 2b. Here is what I did:
A)I found the histogram of each image using matlab and now have represented it using a 1D vector(16 X 16 X 16) . There are 4096 values in a single vector.
B)I generated an ARFF file. It has the following format. There are 1000,000 histograms (1 for each image..thus 1000,000 rows in the file) and 4097 values in each row (image_name + 4096 double values to represent the histogram)
C)The file size is 34 GB. THE BIG QUESTION: HOW THE HECK DO I CLUSTER THIS FILE???
I tried using WEKA and other online tools. But they all hang. Weka gets stuck and says "Reading a file".
I have a RAM of 8 GB on my desktop. I don't have access to any cluster as such. I tried googling but couldn't find anything helpful about clustering large datasets. How do I cluster these entries?
This is what I thought:
Approach One:
Should I do it in batches of 50,000 or something? Like, cluster the first 50,000 entries. Find as many possible clusters call them k1,k2,k3... kn.
Then pick the the next 50,000 and allot them to one of these clusters and so on? Will this be an accurate representation of all the images. Because, clustering is done only on the basis of first 50,000 images!!
Approach Two:
Do the above process using random 50,000 entries?
Any one any inputs?
Thanks!
EDIT 1:
Any clustering algorithm can be used.

Weka isn't your best too for this. I found ELKI to be much more powerful (and faster) when it comes to clustering. The largest I've ran are ~3 million objects in 128 dimensions.
However, note that at this size and dimensionality, your main concern should be result quality.
If you run e.g. k-means, the result will essentially be random because of you using 4096 histogram bins (way too much, in particular with squared euclidean distance).
To get good result, you need to step back an think some more.
What makes two images similar. How can you measure similarity? Verify your similarity measure first.
Which algorithm can use this notion of similarity? Verify the algorithm on a small data set first.
How can the algorithm be scaled up using indexing or parallelism?
In my experience, color histograms worked best on the range of 8 bins for hue x 3 bins for saturation x 3 bins for brightness. Beyond that, the binning is too fine grained. Plus it destroys your similarity measure.
If you run k-means, you gain absolutely nothing by adding more data. It searches for statistical means and adding more data won't find a different mean, but just some more digits of precision. So you may just as well use a sample of just 10k or 100k pictures, and you will get virtually the same results.

Running it several times for independent sets of pictures results in different cluster clusters which are difficult to merge. Thus two similar images are placed in different clusters. I would run the clustering algorithm for a random set of images (as large as possible) and use these cluster definitions to sort all other images.
Alternative: Reduce the compexity of your data, e.g. to a histogram of 1024 double values.

Related

Reducing the file size of MATLAB variables [closed]

Closed. This question needs to be more focused. It is not currently accepting answers.
Want to improve this question? Update the question so it focuses on one problem only by editing this post.
Closed 4 years ago.
Improve this question
Process of extracting data,
I am analyzing 4000 to 8000 DICOM files using matlab codes. DICOM files are read using dicomread() function. Each DICOM file contains 932*128 photon count data coming from 7 detectors. While reading DICOM files, I convert data into double and stored in 7 cell array variables (from seven detectors). So each cell contains 128*128 photon counting data and cell array contain 4000 to 8000 cells.
Question.
When I save each variable separately, size of each variable is 3GB. So for 7 variables it will be 21GB, Saving them and reading back takes awful lot of time. (RAM of my computer is 4GB)
Is there a way to reduce the size of variable?
Thanks.
Different data type will help. You can save data as float instead of double, as DICOM files have it as float too (from http://northstar-www.dartmouth.edu/doc/idl/html_6.2/DICOM_Attributes.html; Graphic Data). This halves size at no loss. You might want to expand to double when doing operations on data to avoid inaccuracies creeping up.
Additional compression by saving it as uint16 (additional x2 space saving) or even uint8 (x4) might be possible, but I would be wary of this - it might work great in all test cases but make problems when you least expect it.
Cell array is not problematic in terms of speed or size - you will not gain (much) by switching to something else. Your data gobbles up memory, not the cell array itself. If you wish, you can save data in a 128x128x7x8000 float array - it should work just fine too.
But if the number of images (this 4000-8000) can increase at any point, rescaling the array will be a pretty costly operation in terms of space and time. Cell arrays are much easier to extend - 8k values to move around instead of 8k*115k=900M values.
Another option is to separate data in chunks. You probably don't need to be working on all 4000 images at once. You can load 500 images, finish your work on them, move on to next 500 images etc. Batch size obviously depends on your hardware and what processing you do with data, but I guess about 500 could be a pretty reasonable starting point.

How to understand Locality Sensitive Hashing? [closed]

Closed. This question does not meet Stack Overflow guidelines. It is not currently accepting answers.
We don’t allow questions seeking recommendations for books, tools, software libraries, and more. You can edit the question so it can be answered with facts and citations.
Closed 2 years ago.
Improve this question
I noticed that LSH seems a good way to find similar items with high-dimension properties.
After reading the paper http://www.slaney.org/malcolm/yahoo/Slaney2008-LSHTutorial.pdf, I'm still confused with those formulas.
Does anyone know a blog or article that explains that the easy way?
The best tutorial I have seen for LSH is in the book: Mining of Massive Datasets.
Check Chapter 3 - Finding Similar Items
http://infolab.stanford.edu/~ullman/mmds/ch3a.pdf
Also I recommend the below slide:
http://www.cs.jhu.edu/%7Evandurme/papers/VanDurmeLallACL10-slides.pdf .
The example in the slide helps me a lot in understanding the hashing for cosine similarity.
I borrow two slides from Benjamin Van Durme & Ashwin Lall, ACL2010 and try to explain the intuitions of LSH Families for Cosine Distance a bit.
In the figure, there are two circles w/ red and yellow colored, representing two two-dimensional data points. We are trying to find their cosine similarity using LSH.
The gray lines are some uniformly randomly picked planes.
Depending on whether the data point locates above or below a gray line, we mark this relation as 0/1.
On the upper-left corner, there are two rows of white/black squares, representing the signature of the two data points respectively. Each square is corresponding to a bit 0(white) or 1(black).
So once you have a pool of planes, you can encode the data points with their location respective to the planes. Imagine that when we have more planes in the pool, the angular difference encoded in the signature is closer to the actual difference. Because only planes that resides between the two points will give the two data different bit value.
Now we look at the signature of the two data points. As in the example, we use only 6 bits(squares) to represent each data. This is the LSH hash for the original data we have.
The hamming distance between the two hashed value is 1, because their signatures only differ by 1 bit.
Considering the length of the signature, we can calculate their angular similarity as shown in the graph.
I have some sample code (just 50 lines) in python here which is using cosine similarity.
https://gist.github.com/94a3d425009be0f94751
Tweets in vector space can be a great example of high dimensional data.
Check out my blog post on applying Locality Sensitive Hashing to tweets to find similar ones.
http://micvog.com/2013/09/08/storm-first-story-detection/
And because one picture is a thousand words check the picture below:
http://micvog.files.wordpress.com/2013/08/lsh1.png
Hope it helps.
#mvogiatzis
Here's a presentation from Stanford that explains it. It made a big difference for me. Part two is more about LSH, but part one covers it as well.
A picture of the overview (There are much more in the slides):
Near Neighbor Search in High Dimensional Data - Part1:
http://www.stanford.edu/class/cs345a/slides/04-highdim.pdf
Near Neighbor Search in High Dimensional Data - Part2:
http://www.stanford.edu/class/cs345a/slides/05-LSH.pdf
LSH is a procedure that takes as input a set of documents/images/objects and outputs a kind of Hash Table.
The indexes of this table contain the documents such that documents that are on the same index are considered similar and those on different indexes are "dissimilar".
Where similar depends on the metric system and also on a threshold similarity s which acts like a global parameter of LSH.
It is up to you to define what the adequate threshold s is for your problem.
It is important to underline that different similarity measures have different implementations of LSH.
In my blog, I tried to thoroughly explain LSH for the cases of minHashing( jaccard similarity measure) and simHashing (cosine distance measure). I hope you find it useful:
https://aerodatablog.wordpress.com/2017/11/29/locality-sensitive-hashing-lsh/
I am a visual person. Here is what works for me as an intuition.
Say each of the things you want to search for approximately are physical objects such as an apple, a cube, a chair.
My intuition for an LSH is that it is similar to take the shadows of these objects. Like if you take the shadow of a 3D cube you get a 2D square-like on a piece of paper, or a 3D sphere will get you a circle-like shadow on a piece of paper.
Eventually, there are many more than three dimensions in a search problem (where each word in a text could be one dimension) but the shadow analogy is still very useful to me.
Now we can efficiently compare strings of bits in software. A fixed length bit string is kinda, more or less, like a line in a single dimension.
So with an LSH, I project the shadows of objects eventually as points (0 or 1) on a single fixed length line/bit string.
The whole trick is to take the shadows such that they still make sense in the lower dimension e.g. they resemble the original object in a good enough way that can be recognized.
A 2D drawing of a cube in perspective tells me this is a cube. But I cannot distinguish easily a 2D square from a 3D cube shadow without perspective: they both looks like a square to me.
How I present my object to the light will determine if I get a good recognizable shadow or not. So I think of a "good" LSH as the one that will turn my objects in front of a light such that their shadow is best recognizable as representing my object.
So to recap: I think of things to index with an LSH as physical objects like a cube, a table, or chair, and I project their shadows in 2D and eventually along a line (a bit string). And a "good" LSH "function" is how I present my objects in front of a light to get an approximately distinguishable shape in the 2D flatland and later my bit string.
Finally when I want to search if an object I have is similar to some objects that I indexed, I take the shadows of this "query" object using the same way to present my object in front of the light (eventually ending up with a bit string too). And now I can compare how similar is that bit string with all my other indexed bit strings which is a proxy for searching for my whole objects if I found a good and recognizable way to present my objects to my light.
As a very short, tldr answer:
An example of locality sensitive hashing could be to first set planes randomly (with a rotation and offset) in your space of inputs to hash, and then to drop your points to hash in the space, and for each plane you measure if the point is above or below it (e.g.: 0 or 1), and the answer is the hash. So points similar in space will have a similar hash if measured with the cosine distance before or after.
You could read this example using scikit-learn: https://github.com/guillaume-chevalier/SGNN-Self-Governing-Neural-Networks-Projection-Layer

Fast spatial data structure for nearest neighbor search amongst non-uniformly sized hyperspheres

Given a k-dimensional continuous (euclidean) space filled with rather unpredictably moving/growing/shrinking  hyperspheres I need to repeatedly find the hypersphere whose surface is nearest to a given coordinate. If some hyperspheres are of the same distance to my coordinate, then the biggest hypersphere wins. (The total count of hyperspheres is guaranteed to stay the same over time.)
My first thought was to use a KDTree but it won't take the hyperspheres' non-uniform volumes into account.
So I looked further and found BVH (Bounding Volume Hierarchies) and BIH (Bounding Interval Hierarchies), which seem to do the trick. At least in 2-/3-dimensional space. However while finding quite a bit of info and visualizations on BVHs I could barely find anything on BIHs.
My basic requirement is a k-dimensional spatial data structure that takes volume into account and is either super fast to build (off-line) or dynamic with barely any unbalancing.
Given my requirements above, which data structure would you go with? Any other ones I didn't even mention?
Edit 1: Forgot to mention: hypershperes are allowed (actually highly expected) to overlap!
Edit 2: Looks like instead of "distance" (and "negative distance" in particular) my described metric matches the power of a point much better.
I'd expect a QuadTree/Octree/generalized to 2^K-tree for your dimensionality of K would do the trick; these recursively partition space, and presumably you can stop when a K-subcube (or K-rectangular brick if the splits aren't even) does not contain a hypersphere, or contains one or more hyperspheres such that partitioning doesn't separate any, or alternatively contains the center of just a single hypersphere (probably easier).
Inserting and deleting entities in such trees is fast, so a hypersphere changing size just causes a delete/insert pair of operations. (I suspect you can optimize this if your sphere size changes by local additional recursive partition if the sphere gets smaller, or local K-block merging if it grows).
I haven't worked with them, but you might also consider binary space partitions. These let you use binary trees instead of k-trees to partition your space. I understand that KDTrees are a special case of this.
But in any case I thought the insertion/deletion algorithms for 2^K trees and/or BSP/KDTrees was well understood and fast. So hypersphere size changes cause deletion/insertion operations but those are fast. So I don't understand your objection to KD-trees.
I think the performance of all these are asymptotically the same.
I would use the R*Tree extension for SQLite. A table would normally have 1 or 2 dimensional data. SQL queries can combine multiple tables to search in higher dimensions.
The formulation with negative distance is a little weird. Distance is positive in geometry, so there may not be much helpful theory to use.
A different formulation that uses only positive distances may be helpful. Read about hyperbolic spaces. This might help to provide ideas for other ways to describe distance.

Most efficient way to store a big DNA sequence?

I want to pack a giant DNA sequence with an iOS app (about 3,000,000,000 base pairs). Each base pair can have a value A, C, T or G. Storing each base pair in one bytes would give a file of 3 GB, which is way too much. :)
Now I though of storing each base pair in two bits (four base pairs per octet), which gives a file of 750 MB. 750 MB is still way too much, even when compressed.
Are there any better file formats for efficiently storing giant base pairs on disk? In memory is not a problem as I read in chunks.
I think you'll have to use two bits per base pair, plus implement compression as described in this paper.
"DNA sequences... are not random; they contain
repeating sections, palindromes, and other features that
could be represented by fewer bits than is required to spell
out the complete sequence in binary...
With the proposed algorithm, sequence will be compressed by 75%
irrespective of the number of repeated or non-repeated
patterns within the sequence."
DNA Compression Using Hash Based Data Structure, International Journal of Information Technology and Knowledge Management
July-December 2010, Volume 2, No. 2, pp. 383-386.
Edit: There is a program called GenCompress which claims to compress DNA sequences efficiently:
http://www1.spms.ntu.edu.sg/~chenxin/GenCompress/
Edit: See also this question on BioStar.
If you don't mind having a complex solution, take a look at this paper or this paper or even this one which is more detailed.
But I think you need to specify better what you're dealing with. Some specifics applications can lead do diferent storage. For example, the last paper I cited deals with lossy compression of DNA...
Base pairs always pair up, so you should only have to store one side of the strand. Now, I doubt that this works if there are certain mutations in the DNA (like a di-Thiamine bond) that cause the opposite strand to not be the exact opposite of the stored strand. Beyond that, I don't think you have many options other than to compress it somehow. But, then again, I'm not a bioinformatics guy, so there might be some pretty sophisticated ways to store a bunch of DNA in a small space. Another idea if it's an iOS app is just putting a reader on the device and reading the sequence from a web service.
Use a diff from a reference genome. From the size (3Gbp) that you post, it looks like you want to include a full human sequences. Since sequences don't differ too much from person to person, you should be able to compress massively by storing only a diff.
Could help a lot. Unless your goal is to store the reference sequence itself. Then you're stuck.
consider this, how many different combinations can you get? out of 4 (i think its about 16 )
actg = 1
atcg = 2
atgc = 3 and so on, so that
you can create an array like [1,2,3] then you can go one step further,
check if 1 is follow by 2, convert 12 to a, 13 = b and so on...
if I understand DNA a bit it means that you cannot get a certain value
as a must be match with c, and t with g or something like that which reduces your options,
so basically you can look for a sequence and give it a something you can also convert back...
You want to look into a 3d space-filling curve. A 3d sfc reduces the 3d complexity to a 1d complexity. It's a little bit like n octree or a r-tree. If you can store your full dna in a sfc you can look for similar tiles in the tree although a sfc is most likely to use with lossy compression. Maybe you can use a block-sorting algorithm like the bwt if you know the size of the tiles and then try an entropy compression like a huffman compression or a golomb code?
You can use the tools like MFCompress, Deliminate,Comrad.These tools provides entropy less than 2.That is for storing each symbol it will take less than 2 bits

Data structure for finding nearby keys with similar bitvalues

I have some data, up to a between a million and a billion records, each which is represented by a bitfield, about 64 bits per key. The bits are independent, you can imagine them basically as random bits.
If I have a test key and I want to find all values in my data with the same key, a hash table will spit those out very easily, in O(1).
What algorithm/data structure would efficiently find all records most similar to the query key? Here similar means that most bits are identical, but a minimal number are allowed to be wrong. This is traditionally measured by Hamming distance., which just counts the number of mismatched bits.
There's two ways this query might be made, one might be by specifying a mismatch rate like "give me a list of all existing keys which have less than 6 bits that differ from my query" or by simply best matches, like "give me a list of the 10,000 keys which have the lowest number of differing bits from my query."
You might be temped to run to k-nearest-neighbor algorithms, but here we're talking about independent bits, so it doesn't seem likely that structures like quadtrees are useful.
The problem can be solved by simple brute force testing a hash table for low numbers of differing bits. If we want to find all keys that differ by one bit from our query, for example, we can enumerate all 64 possible keys and test them all. But this explodes quickly, if we wanted to allow two bits of difference, then we'd have to probe 64*63=4032 times. It gets exponentially worse for higher numbers of bits.
So is there another data structure or strategy that makes this kind of query more efficient?
The database/structure can be preprocessed as much as you like, it's the query speed that should be optimized.
What you want is a BK-Tree. It's a tree that's ideally suited to indexing metric spaces (your problem is one), and supports both nearest-neighbour and distance queries. I wrote an article about it a while ago.
BK-Trees are generally described with reference to text and using levenshtein distance to build the tree, but it's straightforward to write one in terms of binary strings and hamming distance.
This sounds like a good fit for an S-Tree, which is like a hierarchical inverted file. Good resources on this topic include the following papers:
Hierarchical Bitmap Index: An Efficient and Scalable Indexing Technique for Set-Valued Attributes.
Improved Methods for Signature-Tree Construction (2000)
Quote from the first one:
The hierarchical bitmap index efficiently supports dif-
ferent classes of queries, including subset, superset and similarity queries.
Our experiments show that the hierarchical bitmap index outperforms
other set indexing techniques significantly.
These papers include references to other research that you might find useful, such as M-Trees.
Create a binary tree (specifically a trie) representing each key in your start set in the following way: The root node is the empty word, moving down the tree to the left appends a 0 and moving down the right appends a 1. The tree will only have as many leaves as your start set has elements, so the size should stay manageable.
Now you can do a recursive traversal of this tree, allowing at most n "deviations" from the query key in each recursive line of execution, until you have found all of the nodes in the start set which are within that number of deviations.
I'd go with an inverted index, like a search engine. You've basically got a fixed vocabulary of 64 words. Then similarity is measured by hamming distance, instead of cosine similarity like a search engine would want to use. Constructing the index will be slow, but you ought to be able to query it with normal search enginey speeds.
The book Introduction to Information Retrieval covers the efficient construction, storage, compression and querying of inverted indexes.
"Near-optimal hashing algorithms for approximate nearest neighbor in high dimensions", from 2008, seems to be the best result as of then. I won't try to summarize since I read it over a year ago and it's hairy. That's from a page on locality-sensitive hashing, along with an implementation of an earlier version of the scheme. For more general pointers, read up on nearest neighbor search.
This kind of question has been asked before: Fastest way to find most similar string to an input?
The database/structure can be
preprocessed as much as you like
Well...IF that is true. Then all you need is a similarity matrix of your hamming distances. Make the matrix sparse by pruning out large distances. It doesn't get any faster and not that much of a memory hog.
Well, you could insert all of the neighbor keys along with the original key. That would mean that you store (64 choose k) times as much data, for k differing bits, and it will require that you decide k beforehand. Though you could always extend k by brute force querying neighbors, and this will automatically query the neighbors of your neighbors that you inserted. This also gives you a time-space tradeoff: for example, if you accept a 64 x data blowup and 64 times slower you can get two bits of distance.
I haven't completely thought this through, but I have an idea of where I'd start.
You could divide the search space up into a number of buckets where each bucket has a bucket key and the keys in the bucket are the keys that are more similar to this bucket key than any other bucket key. To create the bucket keys, you could randomly generate 64 bit keys and discard any that are too close to any previously created bucket key, or you could work out some algorithm that generates keys that are all dissimilar enough. To find the closest key to a test key, first find the bucket key that is closest, and then test each key in the bucket. (Actually, it's possible, but not likely, for the closest key to be in another bucket - do you need to find the closest key, or would a very close key be good enough?)
If you're ok with doing it probabilistically, I think there's a good way to solve question 2. I assume you have 2^30 data and cutoff and you want to find all points within cutoff distance from test.
One_Try()
1. Generate randomly a 20-bit subset S of 64 bits
2. Ask for a list of elements that agree with test on S (about 2^10 elements)
3. Sort that list by Hamming distance from test
4. Discard the part of list after cutoff
You repeat One_Try as much as you need while merging the lists. The more tries you have, the more points you find. For example, if x is within 5 bits, you'll find it in one try with about (2/3)^5 = 13% probability. Therefore if you repeat 100 tries you find all but roughly 10^{-6} of such x. Total time: 100*(1000*log 1000).
The main advantage of this is that you're able to output answers to question 2 as you proceed, since after the first few tries you'll certainly find everything within distance not more than 3 bits, etc.
If you have many computers, you give each of them several tries, since they are perfectly parallelizable: each computer saves some hash tables in advance.
Data structures for large sets described here: Detecting Near-Duplicates for Web Crawling
or
in memory trie: Judy-arrays at sourceforge.net
Assuming you have to visit each row to test its value (or if you index on the bitfield then each index entry), then you can write the actual test quite efficiently using
A xor B
To find the difference bits, then bit-count the result, using a technique like this.
This effectively gives you the hamming distance.
Since this can compile down to tens of instructions per test, this can run pretty fast.
If you are okay with a randomized algorithm (monte carlo in this case), you can use the minhash.
If the data weren't so sparse, a graph with keys as the vertices and edges linking 'adjacent' (Hamming distance = 1) nodes would probably be very efficient time-wise. The space would be very large though, so in your case, I don't think it would be a worthwhile tradeoff.

Resources