> For the complete documentation index, see [llms.txt](https://docs.visualyze.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.visualyze.ai/robot-ai/how-it-works.md).

# How it works

Visualyze Robot AI uses AutoML and Neural Architecture Search(NAS) to train custom machine learning models on image, text, and video data without writing any code.

### Neural Architecture Search (NAS)

For many machine learning projects, choosing a model is one of the complex processes that requires expertise, time, and money. A significant part of creating a deep learning model is trying out different architectures. This requires specialized skills and is challenging in general; It’s a lot of trial and error and the experimentation itself is time-consuming and expensive.

**NAS** automates the task of finding the suitable neural network architecture by searching for the best neural network architecture for the given data. **It allows us to discover architectures far more complicated than what humans may think to try**, and these architectures can be optimized for particular goals. **NAS** has been used to design networks that are on par or outperform hand-designed architectures.<mark style="color:blue;">\[1]\[2]</mark>

### AutoML

**AutoML** abstracts away all of the complex parts of deep learning - model selection and hyperparameter optimization.&#x20;

<mark style="color:purple;">Visualyze RobotAI</mark> uses a collection of **NAS** presets to find the best model for the given problem. The automated architecture search substantially speeds up the development of new deep learning models as developers do not need to painstakingly evaluate different architectures.

![Machine learning tasks automated by Visualyze Auto ML](https://4267863887-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2Fg5nUBIRa6t8a20g3vSj4%2Fuploads%2FoIZxywQz6xb4Inkx15lR%2Fvisualyze-automl.png?alt=media\&token=e922ffce-2533-47e8-83bc-d1fd81dc5d74)

* \[1]Zoph, Barret; Le, Quoc V. (2016-11-04). "Neural Architecture Search with Reinforcement Learning". [arXiv](https://en.wikipedia.org/wiki/ArXiv_\(identifier\)):[1611.01578](https://arxiv.org/abs/1611.01578) \[[cs.LG](https://arxiv.org/archive/cs.LG)]
* \[2]Zoph, Barret; Vasudevan, Vijay; Shlens, Jonathon; Le, Quoc V. (2017-07-21). "Learning Transferable Architectures for Scalable Image Recognition". [arXiv](https://en.wikipedia.org/wiki/ArXiv_\(identifier\)):[1707.07012](https://arxiv.org/abs/1707.07012) \[[cs.CV](https://arxiv.org/archive/cs.CV)].