Google software engineer and computational artist Zaria Howard explains the concepts behind variational autoencoders (VAE) and generative adversarial networks (GAN), including how combining the two models can overcome the pitfalls of using them individually to generate new images.

Resources:
Experiments with Google → https://goo.gle/3jRGPat
Google Developers ML crash course → https://goo.gle/3yHWD69

Further information:
Convolutional variational autoencoder → https://goo.gle/3g0Kjq9
Intro to autoencoders→ https://goo.gle/3jNUYFH
TF-GAN colabs → https://goo.gle/TFGANColabs
GAN Experiments with Google → https://goo.gle/GAN

Stay tuned for more Tech Bytes → http://goo.gle/TechBytes
Subscribe to Women Techmakers → https://goo.gle/WomenTech

Tech Bytes is a series highlighting Black women engineers and developers at Google who are experts in their field.

Music by Jocelyn Mildner

Image Resources:
Understanding Variational Autoencoders (VAEs) → https://goo.gle/3mbHLcL
Convolutional Variational Autoencoder → https://goo.gle/3g0Kjq9
Progressive growing of GANS for improved quality, stability, and variation →
https://goo.gle/3kZc05V
Overview of GAN Structure → https://goo.gle/3iW5aNa
How to Identify and Diagnose GAN Failure Modes → https://goo.gle/3AUaTsY
What The Heck Are VAE-GANs? → https://goo.gle/3k30Pah

#TechBytes #developer #womenintech

re_ty: Publish;