# Technical software and Artificial Intelligence

### Simulations

Often in engineering, a simulation of a specific system is needed. The first step is always to describe the system using laws of physics, to obtain a continuous mathematical model of the system. Next, this mathematical model is discretized and combined with start and boundary conditions. With this information, numerical algorithms can be put to work to produce the simulation results.

### Planning, optimization, assignment problems

A range of planning, assignment and optimization problems can be solved using linear programming, for instance to calculate a work schedule while taking into account a set of constraints.

### Classification problems

Based on a set of samples and their classes, an artificially intelligent system can be trained to classify unseen samples according to the logic that was found to exist in the known samples. Applications are numerous, for instance in fraud detection or medical research.

### Trend prediction, forecasting

Based on a known series of data points, an artificially intelligent system can be trained to learn the underlying generating function. This is particularly useful in case no trend is visible to the human eye, or when the data is noisy. Once the system is trained, the trend can be extended for a limited number of data points "in the future", effectively predicting the trend. Applications are to be found in financial forecasting.

### Pattern recognition

This is in fact a classification problem. A known set of images, audio or any digital information source together with their classes are fed to an artificially intelligent system. After the system has learned from these examples, it will be able to recognize new images, audio, ... and map it to the right class. One of the applications is the recognition of hand written text, like an address on an envelope.