[HKML] Hong Kong Machine Learning Meetup Season 3 Episode 3

When?

  • Wednesday, October 28, 2020 from 7:00 PM to 9:00 PM (Hong Kong Time)

Where?

  • At your home, on zoom. All meetups will be online as long as this COVID-19 crisis is not over.

Programme:

Talk 1:

Or Lenchner (Luminati Networks) – Why clean, unbiased data is pertinent in every industry today, and how it’s especially critical for AI and machine learning developments

Data-driven insights are critical to multiple industries, including Artificial Intelligence and Machine Learning. AI systems are only as powerful as the information they are built on - clean, unbiased, and gargantuan data is key to training them effectively and keeping them up to date.

In this presentation, Or Lenchner, CEO of leading data collection platform, Luminati Networks, will give an overview of business use cases and how the data collection network can best be leveraged, how to gather enormous amounts of the most accurate and clean data using a proxy network for effective machine learning practices, how data collection can aid in research, and he will also discuss why ethical data collection needs to be paramount.

Talk 2:

Divya Saxena (Hong Kong Polytechnic University) – Generative Adversarial Networks (GANs): Challenges, Solutions, and Future Directions

Dr. Saxena will present her recent review on GANs: Generative Adversarial Networks (GANs): Challenges, Solutions, and Future Directions.

Her slides.

Video Recording of the HKML Meetup on YouTube

  • YouTube videos:

Hong Kong Machine Learning Meetup Season 3 Episode 3 - Talk 1

Hong Kong Machine Learning Meetup Season 3 Episode 3 - Talk 1

Personal Takeaways

The first talk was about collecting clean data at scale. Their proxy network for crawling and scraping as a service seems interesting. I will probably try their solution for the side projects I have which rely on scraped data from the web. If I do so, I may write a tutorial on it.

Luminati Networks is quite big: 180+ employees, PE firm EMK took a USD $200m stake in it.

The second talk was about a review on GANs. The main contribution of this review is a taxonomy of existing GANs. Various GAN improvements can be subdivided into three classes: i) modifications of the neural network architecture; ii) different loss functions; iii) different optimization algorithms.

Thanks to our patrons for supporting the meetup!

Check the patreon page to join the current list: