Machine learning-based data mining techniques (decision trees, bayesian networks, rule based systems, neural networks,….) aim to build a model from past experience (historical data), that later will be used to predict the output of new cases or to get some insight by using the model as a descriptive tool. Greedy algorithms (hill climbing, gradient descent, …) are perhaps the most commonly approach to the design of machine learning algorithms because their good tradeoff between the quality of the obtained model and the amount of resources (mainly CPU time) they need. However, it is also well known that the use of more complex (in terms of resources) approaches often yields more accurate models. Specially the evolutionary algorithms have been used widely in different tasks of Data Mining: classification, clustering, dependence modelling, regression, time series, discovery of comprehensible and interesting knowledge, scaling up for very large databases, etc. Recently other metaheuristics as ant colony optimization, and tabu search among others, are being used in this area.
Dear Maysam,
This will be an answer of an engineer, civil one :-)
There are problems to solve, where according to real numbers and an unknown combination of real numbers solving time is really unlimited. That's why we use heuristic algorithms.
Moreover not every algorithm is suitable to apply in a given problem, so we have the reason why exceptionally clever brains invent new heuristic algorithms. The other way is combining heuristics.
It was proved by many researchers that utilization of one heuristics as a part of an engine of another heuristics significantly improve results - so it is a way of creating meta-heuristic algorithms.
Examples: gradient descent in weights finding in Matlab NN, GA for NN topology finding, a fuzzy output from NN, tabu search for association analysis. These I met, but certainly, there are many other combinations. This is my simple, engineer's understanding of the interplay between heuristics (creating meta-heuristics).
I'm not sure about your statement that "... more complex approaches often yields more accurate models." It seems to me that it is quite easy to overheat with complexity.
What really bothers me is the way of finding inputs that really influence the process we model (when the influence is meaningful, in order no to feed the model with the garbage).
Best Regards
Hubert
In my opinion GA and NN are similar in term of CPU time. Both take a lot of resources :-)
But currently it is not my subject of interest so maybe new NN algorithms are much better. If you interested, see my old papers:
Article Clustering geobotanical data with the use of a genetic algorithm
Article Specialized Measures of Leaf Outlines and Hierarchical Perce...
NLP (Natural language processing) and Machine Learning are both fields in computer science related to AI (Artificial Intelligence). Machine learning can be applied in many different fields. NLP takes care of “understanding” the natural language of the human that the program (e.g. chatbot) is trying to communicate with. This understanding enables the program e.g. chatbot) to both interpret input and produce output in the form of human language.
The machine “learns” and uses its algorithms through supervised and unsupervised learning. Supervised learning means to train the machine to translate the input data into a desired output value. In other words, it assigns an inferred function to the data so that newer examples of data will give the same output for that “learned” interpretation. Unsupervised learning means discovering new patterns in the data without any prior information and training. The machine itself assigns an inferred function to the data through careful analysis and extrapolation of patterns from raw data. The layers are for analyzing the data in an hierarchical way. This is to extract, with hidden layers, the feature through supervised or unsupervised learning. Hidden layers are part of the data processing layers in a neural network.
Featured CBM: Building an IBM Watson Powered AI Chatbot
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Graphs