There has been a large increase in interest in generative AI models of late.

Here’s a great introductory article (complete with code) on GAN’s in TensorFlow.

hese are models that can learn to create data that is similar to data that we give them. The intuition behind this is that if we can get a model to write high-quality news articles for example, then it must have also learned a lot about news articles in general. Or in other words, the model should also have a good internal representation of news articles. We can then hopefully use this representation to help us with other related tasks, such as classifying news articles by topic. Actually training models to create data like this is not easy, but in recent years a number of methods have started to work quite well. One such promising approach is using Generative Adversarial Networks (GANs). The prominent deep learning researcher and director of AI research at Facebook, Yann LeCun, recently cited GANs as being one of the most important new developments in deep learning:

Silicon may be at the heart of most gadgets, but it’s not the only semiconductor around.

Gallium nitride has been getting a lot of attention recently for it’s electrical properties, which outperform silicon in a lot of areas.

Gallium nitride has the potential to revolution power systems, including solar, electric vehicles, and even phone chargers.

Beyond that, it’s finding uses in the mobile industry, and could even be used to build ultra fast processors.

But how feasible is any of that, and even if it’s possible, how long will it take? 

Recently, the researchers from MIT introduced a new AI system known as Timecraft that has the capability to synthesize time-lapse videos depicting how a given painting might have been created.

According to the researchers, there are various possibilities and unique combinations of brushes, strokes, colors, etc. in a painting and the goal behind this research is to learn to capture this rich range of possibilities.

Creating the exact same piece of a famous painting can take days even by skilled artists. However, with the advent of AI and ML, we have witnessed the emergence of a number of AI Artists for a few years now. One of the most popular artisanship of AI is the portrait of Edmond Belamy that was created by Generative Adversarial Network (GAN) and sold for an incredible $432,500.

Text-to-speech engines are usually multi-stage pipelines that transform the signal into many intermediate representations and require supervision at each step.

When trying to train TTS end-to-end, the alignment problem arises: Which text corresponds to which piece of sound?

This paper uses an alignment module to tackle this problem and produces astonishingly good sound.

Paper: https://arxiv.org/abs/2006.03575
Website: https://deepmind.com/research/publications/End-to-End-Adversarial-Text-to-Speech

Content index:

  • 0:00 – Intro & Overview
  • 1:55 – Problems with Text-to-Speech
  • 3:55 – Adversarial Training
  • 5:20 – End-to-End Training
  • 7:20 – Discriminator Architecture
  • 10:40 – Generator Architecture
  • 12:20 – The Alignment Problem
  • 14:40 – Aligner Architecture
  • 24:00 – Spectrogram Prediction Loss
  • 32:30 – Dynamic Time Warping
  • 38:30 – Conclusion

Anyone can simply upload a selfie to the ‘Selfie 2 Waifu’ website to create their own AI-generated waifu-style anime character in seconds.

The AI leverages the release of the new Tensorflow implementation of unsupervised generative network U-GAT-IT, 

For best results, participants are advised to submit a clear passport-style picture with a plain background. Although the site can also generate male anime results, as the name suggests the system generally performs better on female selfies.

“Our model guides the translation to focus on more important regions and ignore minor regions by distinguishing between source and target domains based on the attention map obtained by the auxiliary classifier,” the researchers explain. “These attention maps are embedded into the generator and discriminator to focus on semantically important areas, thus facilitating the shape transformation.”

Deepfakes leverage powerful techniques from machine learning and artificial intelligence to generate visual and audio content with a such a high degree of realism that it has enormous potential to deceive.

This article in Medium explores efforts into research and development into creating countermeasures to such bogus content.

Within recent months, a number of mitigation mechanisms have been proposed and cited with the use of Neural Networks and Artificial Intelligence being at the heart of them. From this, we can distinguish that a proposal for technologies that can automatically detect and assess the integrity of visual media is therefore indispensable and in great need if we wish to fight back against adversarial attacks. (Nguyen, 2019)