33

Well, machine learning in the sense of statistical pattern recognition and data mining are definitely hotter areas, but I wouldn't say research in evolutionary algorithms has particularly slowed. The two areas aren't generally applied to the same types of problems. It's not immediately clear how a data driven approach helps you, for instance, figure out how ...


23

The difference is that in supervised learning the "categories", "classes" or "labels" are known. In unsupervised learning, they are not, and the learning process attempts to find appropriate "categories". In both kinds of learning all parameters are considered to determine which are most appropriate to perform the classification. Whether you chose ...


21

Some decades ago, people thought that genetic and evolutionary algorithms were swiss-army-knives, fueled by spectacular early results. Statements like the building block hypothesis were made in an effort to prove that they were in general good strategies. However, rigorous results were slow in coming and often sobering, most prominently the No Free Lunch ...


13

Note that there are more than 2 degrees of supervision. For example, see the pages 24-25 (6-7) in the PhD thesis of Christian Biemann, Unsupervised and Knowledge-free Natural Language Processing in the Structure Discovery Paradigm, 2007. The thesis identifies 4 degrees: supervised, semi-supervised, weakly-supervised, and unsupervised, and explains the ...


13

A critical part of the story, as I see it, is missing from the other answers so far: Genetic algorithms are mostly useful for brute force search problems. In many contexts, simpler optimization strategies or inference models (what you would broadly call machine learning) can perform very well, and do so far more efficiently than brute force search. ...


13

If you have a neural network (aka a multilayer perceptron) with only an input and an output layer and with no activation function, that is exactly equal to linear regression. In your case, each attribute corresponds to an input node and your network has one output node, which represents the target value you're trying to predict. However, most neural ...


12

To some extent, machine learning is becoming more mathematical and with algorithms able to be 'proven' to work. In some ways, GAs are very "wth happened in there" and you can't perfectly answer the question "so what did your program do?" (well in some people's eyes, anyway). I personally advocate combining neural nets and GA = GANNs. In my honours thesis, ...


10

I suggest a variation of distribution counting: Read the text and insert all the word encountered into a trie, maintaining in each node a count, how often the word represented by this node has occured. Additionally keep track of the highest word count say maxWordCound . -- $O(n)$ Initialize an array of size maxWordCount. Entry type are lists of strings. -- $...


10

You might be interested in learning about grammar induction: given a set of examples of strings from a context-free language, there are algorithms to learn a context-free grammar that generates those strings. To learn more about it, read the Wikipedia article I linked to, and Inducing a context free grammar, Is there a known method for constructing a ...


9

Information retrieval is based on a query - you specify what information you need and it is returned in human understandable form. Information extraction is about structuring unstructured information - given some sources all of the (relevant) information is structured in a form that will be easy for processing. This will not necessary be in human ...


9

While both Operations Research and Data Science both cover a large amount of topics and areas, I'll try to give my perspective on what I see as the most representative and mainstream parts of each. As others have pointed out, the bulk of Operations Research is primarily concerned with making decisions. While there are many different ways to determine how ...


7

You could use any function that gives a lower weight to older entries. For example, if data consists of scores, $s_1,\ldots,s_n$, where the index corresponds to the 'time of arrival' of the entry, that is, newer entries have larger indices, then you could use a weight function that increase as $i$ increases. So any 'increasing' function will do. Examples ...


7

http://gate.ac.uk/ie/ gives a very nice, concise distinction: Information Extraction is not Information Retrieval: Information Extraction differs from traditional techniques in that it does not recover from a collection a subset of documents which are hopefully relevant to a query, based on key-word searching (perhaps augmented by a thesaurus). ...


7

The idea in matrix factorization is to find the latent variables which connect the input and the output. Suppose that we are interested in a movie recommendation system, and that movies "live" on a one-dimensional axis, having romantic comedies in one end, and action movies in the other. Each "input" and "output" move can be rated in this scale (say $1$ to $-...


6

Multiplying by the $n * p$ matrix decreases the dimensionality of the data set. Think of this as projecting the highly dimensional space into a smaller dimensional space. For example, you could do principle component analysis and project it into a small space. This way things that are correlated together are projected into the same dimension and if one of ...


6

It depends entirely on context. In mathematics as a whole, $\log$ usually denotes the natural logarithm (base $\mathrm{e}$). In computer science, the situation isn't as clean because we often want to talk about things like a number of bits or the height of a binary tree and, in those cases, the most natural (*baddum-tsh*) logarithm to use is base-$2$. ...


5

There are tons of tutorials for various backgrounds. What is your background? Here is one list of tutorials: http://www.kernel-machines.org/tutorials


5

if I change a single index the whole file will need to be rewritten So don't change indices. When you add data, remember what words already existed, and store subsequent words afterward. For each word, instead of storing one index in the index file, store several. There is a lot to optimize on top of that. But I don't think this is a productive use of your ...


4

Machine learning unveils a large portion of mathematical apparatus to be developed and applied. Genetics algorithms mostly done by heuristics.


4

In general, the problem of identifying dates and other temporal markers in text is called the problem of extracting temporal references. The search linked will take you to papers related to this.


4

Check out Metacademy. It is a wonderful learning guide for many topics in machine learning. Specifically, this page lists various resources for learning about support vector machines.


4

One of the popular models is the Bag of Words model Also, you can model the words as integers.. they have 'relative distance metrics' for that, and capture the very essence of the classification process. However a downside to that is the preprocessing step is expensive and also you need to have some domain knowledge. A pretty famous distance metric is the ...


4

The state of the art in such problems is done these days via deep neural networks. Among others, two popular and recent approaches for solving the problem of detection and localization of objects are the YOLO paper, and the faster-RCNN, which run a classifier over many variously sized regions in an image. As humans, boats and cars are popular object ...


4

This isn't a full answer, since mhum's is quite good in contrasting the differing aims of OR vs DS. Rather, I want to address this comment of yours: I was wondering if, for example, one could use any OR techniques to solve DS problems. The answer is yes. The clearest example that comes to mind is Support Vector Machines (SVMs). To "fit" an SVM model to ...


3

As far as I know there is no unique definition for what is an outlier/anomaly. Therefore you'll have to decide by yourself what the characteristics of outlying data points in your data set are. This could be for example: distance to the cluster center (threshold), local neighbourhood (a data point that has no/few data points in its neighgbourhood might be an ...


3

I don't think there is anything that should stop you. A couple of tips to set you on the right path: The basic principles of Machine Learning are very simple. They are often formalized in a way that makes it look very complicated, but this is only necessary if you want to get into the details of why things work. The general framework of what makes a ...


3

Decision tree classifiers could be easily converted to rule based classifiers in data mining and vice Versa. please have a look at "Introduction to Data Mining", chapters 4 and 5. Usually it depends on the data characteristics, how to choose between these two methods and which of the mentioned methods give you smaller classification errors.


3

Some background info: You may want to start by studying perceptrons and the learning algorithm (a building block for SVMs). It may also be useful to read up a bit on kernel methods (dealing with high dimensional data) in machine learning and Lagrange multipliers and how to apply them in different optimization tasks. For understanding the theory: There is ...


3

For a formal definition of document in the Information Retrieval context, you can look at IR glossaries. A common definition is: Document: Specific unit of retrieval (usually text). It can be a web page, an article, a book, a section or chapter. (see for example the Glossary of the book Modern Information Retrieval by Ricardo Baeza-Yates and Berthier ...


3

It's hard for me to tell, but it sounds like you want to automatically detect the mammograms where the outline is incorrect. I'm assuming the desired outline is the perimeter of some convex region in the image. I'm not sure whether clustering is going to be the ideal approach here. The first approach that comes to mind for me is something like this: we ...


Only top voted, non community-wiki answers of a minimum length are eligible