Categories of innovation in AI

Gartner suggests that AI innovations in Hype-Cycle down below fall into four categories.

1. Data-centric AI

2. Model-centric AI

3. Applications-centric AI

4. Human-centric AI

Data-centric AI

Unlike traditional AI solutions, Data-centric AI focus on enhancing and enriching the data used to train the algorithms.
Innovations in data-centric AI include synthetic data, knowledge graphs, data labeling and annotation.

Synthetic data

Synthetic data is a class of data that is artificially generated rather than obtained from direct observations of the real world.
Using synthetic data has advantages :

1. Avoids usage of personally identifiable information
2. Reduces cost and save time in ML devenlopment (cheap and faster to obtain)
3. Improves ML performance (increasement of data quantity)

Knowledge graph

As known as semantic network, represents a network of real-world entities and illustrates the relationship between them.
Three main components:

1. Nodes can be ant objects, place, person or things.
2. Edges defines relationship between nodes.
3. Labels labels.