How to create ML models that teach themselves | C2C Community

How to create ML models that teach themselves

  • 14 November 2022
  • 3 replies
  • 23 views

Userlevel 7
Badge +35

Did you ever think about how self-driving cars work?

How do ML and AI make decisions themselves without human touch?

How ML and AI make decisions themself without human touch. One of the biggest challenges is training ML models with appropriate data sets, but what if the ML models could teach themselves? In this episode of Architecting with Google Cloud, Debi Cabrera, Google Cloud developer Advocate chat with Harrison at Helm.ai about how they help companies scale their capabilities with ML models.

Check out this video so the next time you see a self-driving car you’ll know what’s happening under the hood!

 

Click on the below to watch it in detail.

 

Chapter

0:00 - Intro

0:48 - Helm.ai overview

1:37 - Supervised vs unsupervised learning (deep teaching)

3:24 - Real-world example of unsupervised learning

4:50 - Can the software be used on more than cars?

5:47 - Helm.ai architecture

12:14 - Future technical enhancements

14:18 - Wrap Up

 

Architecting with Google Cloud playlist → https://goo.gle/ArchitectingWithGoogl...

Subscribe to Google Cloud Tech → https://goo.gle/GoogleCloudTech


3 replies

Userlevel 7
Badge +65

Hi @malamin,

Thanks for sharing.

But everything must work on the edge to use AI & ML with cars and have them drive by themselves. Otherwise, accidents will happen.

Userlevel 7
Badge +29

AI is already used in the US in some states with cars delivering pizza etc 🙂

Userlevel 7
Badge +35

You’re welcome @ilias ,

This is right to think about how will make decision Self-driving cars about ethical point of view while there is no option to avoid accident. So it is big depute issue the ethical dilemma of self-driving cars.

https://www.ted.com/talks/patrick_lin_the_ethical_dilemma_of_self_driving_cars?language=en

https://www.ted.com/talks/eleanor_nelsen_would_you_sacrifice_one_person_to_save_five?language=en

 

 

Reply