You will find video-tutorials, descriptions und ready made downloadable project files that you can download and import in AIUI on this page.
Browse around and find an AI Solution that will suit your use case.
In this video you can see how AI-UI and your own text can be used to create an AI that tries to predict the next character after 100 characters.
You can play around and try to predict whole words, choose a German text and understand how to unite AI and text fundamentally.
This test project shows how to approach classifying your own texts into certain categories.
These can be categories like “Good | Bad”, or much more complex ones. For example, if you want to pre-filter your emails and automatically classify them into “Technical question | Product question | Spam | Customer feedback”.
There are no limits to your creativity in this respect.
This tutorial shows how to use AI-UI to solve image segmentation problems with AI.
You use your own images and mark the objects that you want an AI algorithm to recognize later. In addition to the balloons shown here, these can also be other objects, such as defective spots on products, people, or any other form of object.
Image recognition thus becomes fully autonomous and fully automatic in no time at all.
If youless complex image recognition tasks to solve and do not want to mark the objects directly pixel by pixel on the image, you can easily classify entire image stacks into classes via AI.
This procedure is ideally suited for IO | nIO classification in production processes or for classifying people, animals, and objects into certain groups.
The following tutorial is about using a .CSV file which contains measured values of orchid flowers in a tabular form.
If a collection of measured values is typical for a certain class, the AI algorithm can be trained effectively and quickly that recognizes the correlations between these values and also evaluates which class it is in the future based on unseen data.
In this “Stock Predictor” example, we would like to show how AI-UI can be used to automatically predict trading recommendations based on past market patterns.
For this purpose, we used historical data of the DAX30 index to generate a time series prediction from an AI. As a result, we obtain a forecast of how likely a buy, hold or sell scenario is.
The goal of all this is to derive predictions or statements about system behavior on the basis of a table of measured values or data in general. This is not only useful in the financial technology sector, but also for predictive maintenance, condition monitoring and quality control.
This four part tutorial deals with the structure and functionality of a self-generated AI for process and/or machine monitoring.
Similar to a digital twin, it is possible to detect system deviations even if they have never occurred in this form before.
The so-called “autoencoder” is often the tool of choice for this purpose. This special form of neural networks also offers other possibilities. Here, we will primarily examine the use case that seems most relevant for Industry 4.0.
In the following tutorial, we show how you can use machine data to evaluate whether a process leads to rejects or not. This article was written in cooperation with the Kompetenzzentrum Mittelstand 4.0 and the Department of Production Engineering at the Technical University of Ilmenau.
https://www.researchgate.net/publication/340224906_In-situ_monitoring_of_hybrid_friction_diffusion_bonded_EN_AW_1050EN_CW_004A_lap_joints_using_artificial_neural_nets
On 30.09.2021 there will be an AI cooking show related to this very project, if you are interested you can find all the information here:
https://www.kompetenzzentrum-ilmenau.digital/veranstaltung/das-erfolgsrezept-fuer-den-einsatz-kuenstlicher-intelligenz-die-ki-kochshow-zur-optimierung-der-produktion/
Disclaimer
Some projects might exceed the possible number of allowed data points or images for the version you are using. Downloading and accessing is possible, but transformations are limited.