Creating an AI Solution Using AI Contents
To solve various industry challenges using artificial intelligence (AI), domain-specific AI Contents are available. These AI Contents are designed to follow the AI Pipeline structure required by ALO, allowing for effective and rapid creation of AI models optimized for the requirements and characteristics of each domain by adjusting the experimental_plan.yaml file.
AI Contents development can encompass the entire process from data collection, preprocessing, model training, evaluation, to deployment. Each step is specified in the experimental_plan.yaml, allowing AI Contents users to apply it to different data, change the model structure, and continuously improve model performance by tuning the model's hyperparameters.
Specifically, the experimental_plan.yaml file contains configurations for various stages of the AI Pipeline, defining data paths, model configurations, training algorithms, and evaluation metrics. This enables experimentation with the most effective data feature extraction, hyperparameter optimization, and model ensemble to discover and refine domain-specific models.
The development and improvement of AI Contents are repeated based on these configuration files, generating and registering AI Solutions tailored to specific industry fields or target problems and moving towards practical operation.
Topics
Exploring AI Contents
Explore the AI Contents page (AI Contents) to select and apply AI Contents suitable for your domain.
Note: For access to the Git code of AI Contents, refer to (AI Contents Access).
| AI Contents | Description |
|---|---|
| TCR | AI Contents for classification/regression of tabular data, automatically performing preprocessing, sampling, HPO, and optimal model selection. https://github.com/mellerikat-aicontents/Tabular-Classification-Regression.git |
| GCR | Similar to TCR in input/output, GCR enhances model accuracy by extracting and utilizing hidden information in data using Graph ML technology, and can handle data with missing values for classification/regression. https://github.com/mellerikat-aicontents/Graph-powered-Classification-Regression.git |
| FCST | AI Contents for analyzing time series data and predicting future values. https://github.com/mellerikat-aicontents/Forecasting.git |
| CV | AI Contents for automating classification of various classes through an image classification model. https://github.com/mellerikat-aicontents/Vision-Classification.git |
| PAD | PAD stands for Point Anomaly Detection, detecting whether point-type time series data is normal or abnormal. https://github.com/mellerikat-aicontents/Anomaly-Detection.git |
| TAD | TAD stands for Tabular AnomalyDetection, which is AI Contents that detects normal and abnormalities through univariate or multivariate variables in the form of a table. https://github.com/mellerikat-aicontents/Tabular-Anomaly-Detection.git |
| MAD | AI Contents for monitoring multiple variables to early detect anomalies or defects (early sensing) and perform anomaly detection. http://mod.lge.com/hub/dxadvtech/aicontents/mad.git |
| VAD | Visual Anomaly Detection (VAD) is an AI content that detects abnormal images by learning normal images. https://github.com/mellerikat-aicontents/Vision-Anomaly-Detection.git |