My Thoughts


Advanced Generative AI (Vision Series): StyleGAN

Sept. 16, 2024 |  Categories:  Tech   Vision   Advanced Vision  

The StyleGAN is a unique attempt by researchers to introduce some structure to the workings of the generator model. It draws from traditional style transfer literature and deals with a learned concept of “style”, which is the focus of the operations within the generator model. Let’s see how this works in practice....

Advanced Generative AI (Vision Series): ProGAN

Aug. 30, 2024 |  Categories:  Tech   Vision   Advanced Vision  

In my previous article, I covered the Vector Quantized Variational Auto Encoder, or the VQ-VAE. In line with my previous series, this time we’ll cover an advanced generative AI model that falls into the family of Generative Adversarial Networks - otherwise known as GANs. This article explores an extension to the traditional GAN known as the Progressive GAN or ProGAN. The ProGAN uses a progressi...

Advanced Generative AI (Vision Series): VQ-VAE

Aug. 15, 2024 |  Categories:  Tech   Advanced Vision  

Originally proposed in 2017, the vector quantized variational autoencoder (VQ-VAE) builds on top of the fundamental backbone of the VAE. If you’d like a review of the concepts behind the VAE, you can check out my earlier article here - for this article I’ll only provide a quick review before going through the theory. The VQ-VAE addresses some fundamental problems with the VAE. First, the VQ-VAE...

Introduction to Generative AI (Vision Series - 3/3): DDPM

July 21, 2024 |  Categories:  Tech   Vision  

Before DDPMs, the primary image generation algorithms were based off the VAE or the GAN. GANs specialize in creating highly realistic images but sometimes struggle to produce a wide range of images. VAEs have the opposite problem - it’s very easy to produce a wide range of images but its harder to get more realistic features with them. Diffusion models are a sort of happy medium between the two...

Introduction to Generative AI (Vision Series - 2/3): GAN

June 30, 2024 |  Categories:  Tech   Vision  

In this series, I'll be dissecting some of the more popular algorithms used for image generation, such as Variational Autoencoders, Generative Adversarial Networks, and Diffusion models. This time, we'll be tackling Generative Adversarial Networks. Originally proposed in 2014, the generative adversarial network (GAN) has become an immensely popular tool for image and data generation. GANs are k...

Introduction to Generative AI (Vision Series - 1/3): VAE

June 28, 2024 |  Categories:  Tech   Vision  

Generative A.I. is currently experiencing an unprecedented advance in popularity and popular use. In this series, I’ll dissect some of the more popular algorithms used for image generation. These algorithms form the backbone for enterprise algorithms like Stable Diffusion and DALLE-3. In particular, the Variational Autoencoder, Generative Adversarial Networks, and Diffusion models have become the ...

Rick and Morty: Modern Satire

Jan. 10, 2024 |  Categories:  Tech  

Animation is a medium that has historically received little love. Up until the release of The Simpsons, animation was largely regarded as an art form solely for children, with adults only hopping onto the trend after its transformation into a medium for political and social commentary. However, recently animation has enjoyed a massive rise in popularity due to the prevalence of shows and media tha...

The Mind and AI

Jan. 10, 2024 |  Categories:  Tech  

For sure, the human brain is remarkable. It’s the most advanced biological construct we know of. But even with all our tools and research and drive, there is still so much we don’t know about the brain. The amount we don’t know about it surprisingly presents serious issues for the future of robotics and artificial intelligence....