#003 | Everything is a System
Hi, I’m Ryan! You are receiving this because you have signed up to my weekly newsletter for Natural Language Processing (#NLP365), Entrepreneurship, and Life Design content!
Hey friends,
In this week’s newsletter, I want to share an important concept with you and that’s the concept of everything is a system.
Everything in life is a system. That means anything in life can be decompose into components that allows you to truly understand how the system works. Once you gained a good understanding of the different components within a system, your ability to operate that system automatically improves as you can CONSCIOUSLY tweak things in your favour.
Having trouble loosing weight? Well, there is a system for that. One of the main components is calories out must be greater than calories in. How do we measure that? First, find out what your maintenance calories is, which you can use the many fitness calculators online to get a starting point. Then, track what you eat daily using fitness apps and make sure that what you consume is below your maintenance calories. Repeat and iterate. Done. You will now lose weight GUARANTEED. I have used this simple component (and others) over the last 5.5 years to alternate between winter bulking and summer shredding seasons. Check out my results here.
Having trouble getting good grades? Well, there is a system for that. The two main components are Active Recall and Spaced Repetition. With these two components, you can absorb more information and get better grades with LESS TIME.
Having trouble getting a job? Well, there is a system for that. One of the components is called the WOW factor.
Having trouble figuring out yourself? Well, there is a system for that. Okay… this one, I am still trying to figure out. It’s a complex one…
Once you figured out what makes up a system, you can then act accordingly and strategise on how to best act to achieve your goals! Essentially, you can hack the system 😈
This week I finished reading:
The Hard Thing about Hard Things (18th Jan - 22nd Jan 2021)
Total: 6 / 26 books | 0 / 26 level 4 notes | 0 / 12 actions
❓Question of the Week
What systems have you figured out that has impacted you the most in your life?
For me, fitness was the first system I cracked. Afterwards, I became obsess with cracking more systems! The biggest and hardest system to crack, IMO, is life.
What do you want to achieve in life? What’s the kind of person you would be proud of when you are 100 years old (I assume everyone dies at 100 years old haha)? What’s the kind of life you would be proud of to have lived when you are 100 years old?
Those were my first set of questions towards figuring out the life system. Check out my Game of Life video for what I got so far!
🐦 Tweet of the Week
It’s common for people to think that discipline and systems restrict your freedom but actually it’s quite the opposite.
Discipline with your daily good habits and over time you will build skills that gets you what you want. Discipline with your morning routine and you will have more time to achieve your goals. Discipline eventually gives you 100% ownership of your own time.
💡 Quote of the Week
Innovation requires a combination of knowledge, skill, and courage — The Hard Thing about Hard Things
🔥 Recommendation(s) of the Week
Audible — I signed up to the Audible 24 credits annual subscription this week! So excited to start buying new audiobooks and listen to more books! Making reading a habit has been one of the high return decision I made this year and it’s only the 3rd week of 2021 🤣
🔦 AI Research Papers Spotlight of the Week
A Survey on Deep Learning for Named Entity Recognition
This paper was a good read to map out and understand the current research landscape of Named Entity Recognition (NER). Methodologies for NER are categorised as follows:
Rule-based
Unsupervised Learning
Feature-based Supervised Learning
Deep Learning for NER
In terms of current challenges in the NER space, there are three:
Data Annotation
Deep learning requires large annotated data. We need high quality and consistency data due to language ambiguity. For each type of entity, we need a certain number of training examples to help the model generalise. The inconsistency in data annotation across datasets and domains means that models trained on one dataset seldom do well in another.
We also need to develop an annotation schemes to account for NESTED entities AND fine-grained entities, where one named entity has multiple types.
NER on Informal Text
NER is still bad on user-generated text due to shortness and noisiness.
Unseen Entities
The robustness and effectiveness of NER system is how they can identify unusual, unseen entities in the context of emerging discussions.
🎥 This Week on YouTube
Episode #005 of Research Papers Summary! This week’s on LUKE Transformer, which introduces a simple yet effective extension to both the pretraining objective and self-attention mechanism to achieve SOTA results in 5 entity-related tasks. Check out the original paper here and my video summary below 🎮
That’s it for this week! I hope you find something useful from this newsletter. More to come next Sunday! Have a good week ahead! 🎮
Create your profile
Only paid subscribers can comment on this post
Check your email
For your security, we need to re-authenticate you.
Click the link we sent to , or click here to sign in.