Tuesday, September 29, 2020

FAME 1

This is the first of a series of posts documenting my experience and thoughts in undertaking the fast.ai machine learning course. And hence, the title: Fast AI ML Experience (FAME).

 

Learning Approach

The following influential excerpt is plucked from the first topic of the fast.ai machine learning course. I was really humbled and inspired by how succinct they describe an underrated, informal (at least for me) approach  to learning or mastering a skill. I always used to think that if I'm gonna be the best, I need to do things by the books. Every fine details need to be covered. That takes the soul (fun) out of the learning. I will add more description when I find more words to express this monumental feeling.


"

I'll be using a top-down teaching method, which is different from how most math courses operate. Typically, in a bottom-up approach, you first learn all the separate components you will be using, and then you gradually build them up into more complex structures. The problems with this are that students often lose motivation, don't have a sense of the "big picture", and don't know what they'll need.

If you took the fast.ai deep learning course, that is what we used. You can hear more about my teaching philosophy in this blog post or in this talk.

Harvard Professor David Perkins has a book, Making Learning Whole in which he uses baseball as an analogy. We don't require kids to memorize all the rules of baseball and understand all the technical details before we let them play the game. Rather, they start playing with a just general sense of it, and then gradually learn more rules/details as time goes on.

All that to say, don't worry if you don't understand everything at first! You're not supposed to. We will start using some "black boxes" such as random forests that haven't yet been explained in detail, and then we'll dig into the lower level details later.

To start, focus on what things DO, not what they ARE.

Your practice

People learn by:

  1. doing (coding and building)
  2. explaining what they've learned (by writing or helping others)

Therefore, we suggest that you practice these skills on Kaggle by:

  1. Entering competitions (doing)
  2. Creating Kaggle kernels (explaining)

It's OK if you don't get good competition ranks or any kernel votes at first - that's totally normal! Just try to keep improving every day, and you'll see the results over time.

To get better at technical writing, study the top ranked Kaggle kernels from past competitions, and read posts from well-regarded technical bloggers. Some good role models include:

"

No comments:
Labels:

Friday, September 4, 2020

September Highlights

September's the time for stepping up & step them bears! Here we go! 

 

1. Problems, problems everywhere...

Problem statements from Malaysia Cross Industry Virtual Hackathon 👀

Celcom: Telecommunications  

1. For Oil & Gas client

How might we digitalise the communication process between employees in an off-shore oil rig and experts that are in the on-shore HQ to ensure that real-time information can be communicated accurately?

2. For Grocery client

How might we improve the supply chain process for the client to ensure that products delivered to customers are fresh & high quality especially when the orders are placed via an e-commerce platform?

3. For Agriculture client

How might we identify which plants are prone to bacterial & fungal infections and develop ways to prevent the diseases from spreading to other crops?

 

Malaysia Airlines: Travel

1. Health & Safety focused

How might we use technology to address the health & safety concerns of passengers during the travel journey?

 2. Travel Experience focused

How might we utilise technology to enhance Malaysia Airlines service to deliver the warmth of Malaysian hospitality?

3. Sustainability focused

How might we engage & collaborate with our passengers to reduce waste on our flights?

 

Ecoworld: Real Estate

How might we develop a single platform to collect insights & feedback from various engagement channels to allow us to serve & manage customer experience?

 

HLX: Retail/Office space focused

How might we increase space utilisation of an office building in the city with low tenancy, especially in this era of changing work practices?


 

 2. That's a heck of a linear regression 😲

 

3. You don't just study something & be an expert. You need to do it.

Entrepreneurship has somehow become the “easy” option to fast-track their way to success, or so they thought. Many are offering “instant” educational programs, both informally through seminars and workshops or formally through university degrees with the promise of making them entrepreneurs. What educators often miss is that there is no research evidence that suggests the possibility of “teaching” someone to become an entrepreneur. You cannot call a subject “entrepreneurship”, and expect that after taking a few classes studying and reading about it, someone will be magically ready to become an entrepreneur.

This article was published in thejakartapost.com with the title "(Ugly) truth about entrepreneurship". Click to read: https://www.thejakartapost.com/academia/2019/12/05/ugly-truth-about-entrepreneurship.html.


Download The Jakarta Post app for easier and faster news access:
Android: http://bit.ly/tjp-android
iOS: http://bit.ly/tjp-io 

 

 

Source: https://www.thejakartapost.com/academia/2019/12/05/ugly-truth-about-entrepreneurship.html  



4. Hilarious




 

5. Laptop for deep learning. Woah, it's very humbling.


Store: https://lambdalabs.com/about?ref=blog

Blogs (Got great content): https://lambdalabs.com/blog/  

 

 

6. Yeeaaa! Yeeaaa! Look who's in the news...

 

https://www.nst.com.my/lifestyle/bots/2020/09/623133/tech-two-utm-students-become-winners-asean-data-science-competition 



7. Math for Machine Learning (from AWS Training)



8. Problem statements from Startup Weekend Singapore 2020: Tomorrow
25-27 September 2020 (Fri-Sun)









9. Maxis IOT Challenge 2020





10. Honest Bee offerings as  modern shop

https://habitat.honestbee.sg/

 

11. Philly Codefest 2020 Theme: Environmental Sustainability 


 

 

12. Estonia is like Europe's Singapore - digital country.

https://e-estonia.com/solutions/e-governance/i-voting/

Education & Research

  • 1st in Europe in the OECD PISA test
  • 100% of schools use e-school solutions
  • every 10th student sets out to study IT every year (twice as many students pursue IT careers in Estonia than the average in other OECD countries.)
  • Higher education is free in Estonia at public universities. 
  • Factors: e-schools, information system for schools & research 

 

 

13. Some new perspectives of linear algebra.

https://www.math.unl.edu/~tshores1/linalg.gif 

https://www.math.unl.edu/~tshores1/linalgtext.html 

 

14. Inokyo (a cheaper alternative to Amazon Go shop automation) 

https://techcrunch.com/2018/08/16/inokyo/ 

How Inokyo works:

1. Download its app and connect a payment method
 
2. Get a QR code that you wave in front of a little sensor as you stroll into the shop
 
3. Overhead cameras will scan your body shape and clothing without facial recognition in order to track you as you move around the store. On-shelf cameras track when products are picked up or put back.
Combined, knowing who’s where and what’s grabbed lets it assign the items to your cart.
 
4. Scan again on your way out, and later you get a receipt detailing the charges. And if you’re overcharged, there’s an in-app support button for getting a refund.
 

Challenges:
- Originally, Inokyo actually didn’t make you scan on the way out, but it got the feedback that customers were scared they were actually stealing. The scan-out is more about peace of mind than engineering necessity.
 
- When a rush of customers floods in and cameras have to differentiate between multiple similar-looking people
 
- These stores will be judged by the convenience of the holistic experience, not just the tech.  
 
-  Human can check scanning, handle restocking, customer support and store maintenance once freed from cashier duties.
 
- The auto-shop race’s winner will be determined by who builds the most accurate tracking software, easy-to-install hardware and pleasant overall shopping flow.
 
 
Standard Cognition
 
Cameras only to see what people are picking up off the shelves and eventually purchasing. 
 
The end goal is for someone to walk in, grab an item off a shelf, and walk out without having to wait in line to pay for it. 


 
 
 
 
 
 


15. Omfg
 
Adesanya = Experience + Actual Learning + Precision + Expression + Self belief  
 
Costa @the_shampion = Do what got you to the dance


Powwerrr + Learning from mistakes + Not repeating mistakes




16. Information overload & scattering > Miss out on full/targeted benefits > Recommender + Congregator + Updates + Links







17. Evernote & ClickUp have some good templates of documenting progress, goals, notes, tasks. Can look them to see how they think & organize these kinds of information.



18. Good video on logical, rational motivation on effort & goals.
 

 

19. Friggin dust




20. Corona to restore the "fear in God" among Klangites.
 
 
 
21. Doing experiments using oneself as the test subject
 
 
22. Bottom Up first, Top Down later or occasionally
 
 
 
Top down learning in simple terms

 
Carried by the winds of interest


 
Feynmann is spoken of here too. Invent something to understand it.




23. One of the most easily forgotten things in the world: No free lunch


24. Some nice recommendations from this blog series

Tensorflow for poets
RNN for Artists

Make Your Own Neural Network by Tariq Rashid. It also has some typos, but there is a great Github repo with updates that fixes most of them. 
 
 
25.  How to study math?
 
Two kinds of understanding a math text: 
- being able to follow, reconstruct, etc. a proof or solution method
- rewiring one’s brain so that it’s obvious why the theorem is true” 
  or why the solution works. 
The second kind is the definite goal.


DONT'S
Don't memorise too much. Keep formulas in a cheatsheet.
Don't be afraid/discouraged to go back & relearn material.

DO
Divide & conquer. Break into smaller manageable blocks.
Skim for overview
Then reread with concentration
Taking notes on separate paper - help keep focus
Do (not read) each step of each example
Think about what you've read.

Fit it in with what you already know. 
  • More general version of a specific case you learned earlier? 
  • Shorter method for a previously longer way? 
  • Does this use something you learned previously that looked useless at the time?
Explain it in your own words without stumbling. No "sort of get it".



25. In many ways, you can approach math like sports:

Deal with cramps right away. When there’s something you don’t understand, you may be tempted to just put it aside and hope for the best. That strategy doesn’t work at all in math! Because everything builds, if you don’t understand A you will probably not understand B and C either.

If this happens in class, ask a question right away. Don’t apologize and don’t worry about looking stupid; probably other people have exactly the same question.

Warm up before the event. Before class, look back over the readings and your homework. Make sure you are ready with any questions.

Stay in training. Review your notes after class, even rewrite them to make sure you understand everything. If there are several days between classes, review the material at least every other day to keep it fresh in your mind.

Make sure you train enough. The College recommends 2–3 hours per classroom hour, but you may need more.
 
 
26. DC showing how to get out of an otherwise sticky ego war.



27. AI Blogs

28.  
 
29. 
 
 
 
 






30. https://www.zdnet.com/article/ai-aint-no-a-student-deepmind-flunks-high-school-math/

 



 
How does a country get money to build infrastructure? Like where does the money come from? If we knew, we can better allocate our time & energy to get it

Entrepreneurship has somehow become the “easy” option to fast-track their way to success, or so they thought. Many are offering “instant” educational programs, both informally through seminars and workshops or formally through university degrees with the promise of making them entrepreneurs. What educators often miss is that there is no research evidence that suggests the possibility of “teaching” someone to become an entrepreneur. You cannot call a subject “entrepreneurship”, and expect that after taking a few classes studying and reading about it, someone will be magically ready to become an entrepreneur.

This article was published in thejakartapost.com with the title "(Ugly) truth about entrepreneurship". Click to read: https://www.thejakartapost.com/academia/2019/12/05/ugly-truth-about-entrepreneurship.html.


Download The Jakarta Post app for easier and faster news access:
Android: http://bit.ly/tjp-android
iOS: http://bit.ly/tjp-ios

How 
No comments:
Labels:
Subscribe to: Posts (Atom)