Uizard Research

Explore the research and technology that makes Uizard possible.

The scientific contributions listed on this page showcase some of the technology powering Uizard. The presented methods, experiments, and results are published for educational purposes only.

We share the science, but we keep some secrets. 😉

Our research was featured in
The Next Web
Wired
Hacker News
Nvidia
Fast Company

March 27 2021

Improving UI Layout Understanding with Hierarchical Positional Encodings

How can we modify transformers for UI-centric tasks? Layout Understanding is a sub-field of AI that enables machines to better process the semantics and information within layouts such as user interfaces (UIs), text documents, forms, presentation slides, graphic design compositions, etc. Companies already invest extensive resources into their web/mobile applications’ UI and user experience (UX), with Fast Company reporting that every $1 spent on UX can return between 2–100x on its investment.

February 17 2021

Generating Design Systems using Deep Learning

How do you enable people to easily design beautiful mobile and web apps when they have no design background? Easy: you just throw some neural networks at the problem! Well, kinda...

September 4 2020

From Research to Production with Deep Semi-Supervised Learning

Semi-Supervised Learning algorithms have shown great potential in training regimes when access to labeled data is scarce but access to unlabeled data is plentiful. In this post, we discuss how Semi-supervisedlearning approaches can be useful for machine learning production environmentsand the lessons we’ve learned using them at Uizard.

December 18 2019

RealMix: Towards Realistic Semi-Supervised Deep Learning Algorithms

Semi-Supervised Learning (SSL) algorithms have shown great potential in training regimes when access to labeled data is scarce but access to unlabeled data is plentiful. However, our experiments illustrate several shortcomings that prior SSL algorithms suffer from. In particular, poor performance when unlabeled and labeled data distributions differ. To address these observations, we develop RealMix, which achieves state-of-the-art results on standard benchmark datasets across different labeled and unlabeled set sizes while overcoming the aforementioned challenges.

June 5 2018

Code2Pix: Deep Learning Compiler for Graphical User Interfaces

This work is a collaboration between Uizard and UC Berkeley’s Statistics Undergraduate Student Association. We provide a deep learning solution to the problem of generating Graphical User Interfaces (GUIs) from a textual description. Additionally, intermediary work hints at the possibility of a single textual to GUI renderer that works across multiple platforms.

November 30 2017

Teaching Machines to Understand User Interfaces

Semi-Supervised Learning algorithms have shown great potential in training regimes when access to labeled data is scarce but access to unlabeled data is plentiful. In this post, we discuss how Semi-supervisedlearning approaches can be useful for machine learning production environmentsand the lessons we’ve learned using them at Uizard.

May 23 2017

pix2code: Generating Code from a Graphical User Interface Screenshot

Semi-Supervised Learning algorithms have shown great potential in training regimes when access to labeled data is scarce but access to unlabeled data is plentiful. In this post, we discuss how Semi-supervisedlearning approaches can be useful for machine learning production environmentsand the lessons we’ve learned using them at Uizard.

Interested in conducting machine learning research with us?