Leveling Up This Week

Had a good well-rounded week here in Silicon Valley.  I’m pretty conscious each week to not neglect my career, relationships, and physical/mental health.  I have a hard to fix habit of both aggressively scheduling my accomplishments for the day/week, and also stumbling onto new and improved things to do.  Overall though if I can slightly net improve for the week I’m happy.

Career wise my main focus is to continuously be better at Machine Learning/AI, as well as learn about new technologies such as cryptocurrencies.  I updated a project that I had built in the past using neural networks (a deep learning algorithm) to predict the price of a stock given other correlated stock prices.  I enjoy the markets because you can immediately test your models and algorithms in the most competitive human/machine arena that we have.  I was going to open source and freely give away my model as I have found almost zero good examples of neural networks and trading on the internet.  There are a few, but they have way less potential than the simple model that I built and tweaked.  I guess the thing stopping me is just the simple fear that I’ll have a stronger motivation to improve and apply it in the future, and I’d be giving away some semi-valuable intellectual property.  At some point though I’d rather someone make else money and improve the markets than it just collect dust inside my computer.

One really cool brand new AI research paper by DeepMind (now owned by Google, they developed a groundbreaking AI program to beat the world’s best Go players which was thought to have little chance until it won last year) has been broken down into a 4 minute video here: https://www.youtube.com/watch?v=vzg5Qe0pTKk

AI is nowhere near general purpose intelligence, and there’s usually just slight tweaks that improve things, but this seems like a pretty huge step forward.  The paper shows how you can combine two types of regularly used algorithms, Convolutional Neural Networks (CNN), Long-Short-Term-Networks (LSTM), and a third one I had never heard of called Relational Networks (RN).  To simplify things the CNN is good at understanding images, the LSTM is good at understanding text, and the RN is good at understanding relationships.  Often you would just use one of these algorithms to do something, but this combination allows a program to look at a picture of something like three yellow balls and two green ones, and answer questions such as “are there more green balls or yellow ones? are the green balls above or below the yellow?” and answer the questions with human level accuracy!  It’s a really interesting field, and research (and people have typed out the code for you to copy and modify) like this is out there free for anyone to use and apply to anything they want.  This is less than a month old, and certainly 99.99% of businesses that could apply this haven’t.

Relationship wise I talked with my mom, brother, girlfriend (whom I would normally see but is out of town), and some of my best friends.  Sometimes I get a call in the middle of my work/studies that decreases my productivity a bit, but if my work isn’t time sensitive I like being available to the people I care about.  When people are always busy and hard to get in touch with it discourages me from contacting them, and there have been periods in my life when I’ve been hard to talk to.  My natural instincts are to be a workhorse and shutout everything that isn’t related to my work, even in periods when I have the time.  I think like most people after my health (I can’t be a good family member/friend if I’m sick) my relationships are definitely the thing I value the most.  I’m regularly reminding myself to touch base with my brother, sister, etc. even if it’s just a text as I appreciate so much when they do the same.  

I’m also trying to provide some mentorship when people reach out for advice.  35 years finally feels old enough to kick back some of the wisdom when people request it, and I’m also just barely wise enough to not give it if it’s not wanted 🙂   One of the advantages of jumping into so many careers is that I have put my brain deep into some pretty diverse fields, and it’s nice to give back.

Physical/mental wise was a good week.  My current fitness plan involves me going to open gym and doing this semi-advanced bodyweight fitness program 3 days per week:

http://antranik.org/rings-routine/

I’m used to CrossFit intensity, but after tweaking my back two weeks ago (too much traveling, drinking at the traveling events, and not sleeping properly) I decided to step back and rebuild my core strength and flexibility for a little bit.  On my off days if I feel good (and sometimes even when I don’t) I do yoga.  I alternate between just following along on free youtube videos in my house, and going to a local yoga studio.  I’ve been more of a talker for many years about how good yoga is, but now I’m finally in the right stage in my life where I actually do it too.  It just makes me feel so good, and my main priority is long-term health and mobility vs. being ripped for the weekend.

Mental health wise I try to spend some regular time in silence.  With all there is to learn/do/coffee! I’m as overstimulated as anyone, so I try to spend time either meditating, sitting outside, or walking around whenever I feel like I need it which turns out to be often.  I also try to read something uplifting every day.  My favorite in the morning is http://greatday.com/ , and in the evening it’s usually a short passage from something on my kindle.  I just have to be careful sometimes as I get motivated and fired up right before I’m supposed to sleep.

Usually I do something outdoorsy, but today I’m in a city vibe, and going to leave you with a New York City picture I took when attending a friend’s birthday a few months ago.