Summary of “Why are Machine Learning Projects so Hard to Manage?”

One constant is that machine learning teams have a hard time setting goals and setting expectations.
In the first week, the accuracy went from 35% to 65% percent but then over the next several months it never got above 68%. 68% accuracy was clearly the limit on the data with the best most up-to-date machine learning techniques.
My friend Pete Skomoroch was recently telling me how frustrating it was to do engineering standups as a data scientist working on machine learning.
Engineering projects generally move forward, but machine learning projects can completely stall.
Machine learning generally works well as long as you have lots of training data *and* the data you’re running on in production looks a lot like your training data.
Machine Learning requires lots and lots of relevant training data.
What’s Next?‍.The original goal of machine learning was mostly around smart decision making, but more and more we are trying to put machine learning into products we use.
As we start to rely more and more on machine learning algorithms, machine learning becomes an engineering discipline as much as a research topic.

The orginal article.

Summary of “Don’t Know What You Want? Improve These 7 Universal Skills”

What does success look like? What do you want from life? What career do you want?
We think it’s the worst thing in the world if you don’t know what you want to do in life.
One of the biggest thinking errors that I’ve made was that I thought I needed to know what I exactly wanted to do with my life.
The truth is that no one knows what they truly want.
So it’s not important to know exactly what you want to do with your life.
It’s not even realistic to boldly claim “I know what I want!”.
If you can’t decide what direction you want to go in life, that’s automatically your #1 goal in life – to figure out where you want to go.
Persuasion: Learn how to get what you want in an ethical way.

The orginal article.

Summary of “The Play Deficit”

In a book called The Play of Animals, Groos argued that play came about by natural selection as a means to ensure that animals would practise the skills they need in order to survive and reproduce.
It explains why young animals play more than older ones and why those animals that depend least on rigid instincts for survival, and most on learning, play the most.
Lion cubs and other young predators play at stalking and pouncing or chasing, while zebra colts and other prey species play at fleeing and dodging.
Groos followed The Play of Animals with a second book, The Play of Man, in which he extended his insights about animal play to humans.
In hunter-gatherer bands, at Sudbury Valley School, and everywhere that children have regular access to other children, most play is social play.
Preschoolers playing a game of ‘house’ spend more time figuring out how to play than actually playing.
Social play is by far the most effective venue for learning such lessons, and I suspect that children’s strong drive for such play came about, in evolution, primarily for that purpose.
We think of play as childish, but to the child, play is the experience of being like an adult: being self-controlled and responsible.

The orginal article.

Summary of “The state of AI in 2019”

Experts refer to this specific instance of AI as artificial general intelligence, and if we do ever create something like this, it’ll likely to be a long way in the future.
No one is helped by exaggerating the intelligence or capabilities of AI systems.
It’s better to talk about “Machine learning” rather than AI. This is a subfield of artificial intelligence, and one that encompasses pretty much all the methods having the biggest impact on the world right now.
How does machine learning work? Over the past few years, I’ve read and watched dozens of explanations, and the distinction I’ve found most useful is right there in the name: machine learning is all about enabling computers to learn on their own.
If you’re not explicitly teaching the computer, how do you know how it’s making its decisions? Machine learning systems can’t explain their thinking, and that means your algorithm could be performing well for the wrong reasons.
Teaching computers to learn for themselves is a brilliant shortcut – and like all shortcuts, it involves cutting corners Teaching computers to learn for themselves is a brilliant shortcut.
There’s intelligence in AI systems, if you want to call it that.
Kai-Fu Lee, a venture capitalist and former AI research, describes the current moment as the “Age of implementation” – one where the technology starts “Spilling out of the lab and into the world.” Benedict Evans, another VC strategist, compares machine learning to relational databases, a type of enterprise software that made fortunes in the ’90s and revolutionized whole industries, but that’s so mundane your eyes probably glazed over just reading those two words.

The orginal article.

Summary of “Scientific American Blog Network”

My first semester I took atomic physics with Professor Delroy Baugh, self-proclaimed “Laser Guy.” I’d never taken a physics course before in my life, and as a reward for my willingness to transcend my comfort zone I received a D. Somewhere between the two of us lies a sweet spot: if you only ever get 100 percent on your tests, they aren’t hard enough.
If you never get above 50 percent, you’re probably in the wrong major.
An article from a team led by University of Arizona cognitive scientist Robert Wilson provides an answer: 15 percent.
The researchers argue that a test is optimally difficult if the test-taker gets 85 percent of the questions right, with 15 percent incorrect.
Under loose assumptions, they show that the optimal error rate for training a broad class of deep learning algorithms is 15 percent.
If you’re a teacher, your tests should be difficult enough that the average score is 85 percent.
Outside the classroom, the implications of the 85 percent rule are similar.
If you are learning a new language, say on Duolingo, then you should be getting about 15 percent of the answers wrong.

The orginal article.

Summary of “The Power of Imperfect Starts”

What is optimal for them right now isn’t necessarily needed for you to get started.
If you set your bar at “Amazing,” it’s awfully difficult to start.
Comparing your current situation to someone who is already successful can often make you feel like you lack the required resources to get started at all.
You don’t need new cooking bowls to start eating healthy.
You don’t need a new backpack to start traveling.
You can point out how your business mentor is successful because they use XYZ software, but they probably got started without it.
Don’t let visions of what is optimal prevent you from getting started in the first place.
An imperfect start can always be improved, but obsessing over a perfect plan will never take you anywhere on its own.

The orginal article.

Summary of “The Secret Algorithm Behind Learning”

Feynman stumbled upon a formula for learning that ensured he understood something better than everyone else.
Not your smart adult friend but rather an 8-year-old who has just enough vocabulary and attention span to understand basic concepts and relationships.
A lot of people tend to use complicated vocabulary and jargon to mask when they don’t understand something.
The problem is we only fool ourselves because we don’t know that we don’t understand.
If you struggle, you have a clear understanding of where you have some gaps.
Identifying the boundaries of your understanding also limits the mistakes you’re liable to make and increases your chance of success when applying knowledge.
If the explanation isn’t simple or sounds confusing that’s a good indication that your understanding in that area still needs some work.
If you really want to be sure of your understanding, run it past someone.

The orginal article.

Summary of “AI predictions for 2019 from Yann LeCun, Hilary Mason, Andrew Ng, and Rumman Chowdhury”

Public awareness of AI still isn’t where she thinks it needs to be and in the year ahead Chowdhury hopes to see more people take advantage of educational resources to understand AI systems and be able to intelligently question AI decisions.
He’s the cofounder of Google Brain, an initiative to spread AI throughout Google’s many products, and the founder of Landing AI, a company that helps businesses integrate AI into their operations.
After more than three years there, in 2017 he left his post as chief AI scientist for Baidu, another tech giant that he helped transform into an AI company.
Ng spoke with VentureBeat earlier this month when he released the AI Transformation Playbook, a short read about how companies can unlock the positive impacts of artificial intelligence for their own companies.
One major area of progress or change he expects to see in 2019 is AI being used in applications outside of tech or software companies.
“I think a lot of the stories to be told next year will be in AI applications outside the software industry. As an industry, we’ve done a decent job helping companies like Google and Baidu but also Facebook and Microsoft – which I have nothing to do with – but even companies like Square and Airbnb, Pinterest, are starting to use some AI capabilities. I think the next massive wave of value creation will be when you can get a manufacturing company or agriculture devices company or a health care company to develop dozens of AI solutions to help their businesses.”
Yann LeCun is a professor at New York University, Facebook chief AI scientist, and founding director of Facebook AI Research, a division of the company that created PyTorch 1.0 and Caffe2, as well as a number of AI systems – like the text translation AI tools Facebook uses billions of times a day or advanced reinforcement learning systems that play Go. LeCun believes the open source policy FAIR adopts for its research and tools has helped nudge other large tech companies to do the same, something he believes has moved the AI field forward as a whole.
The democratization of AI, or expansion to corners of a company beyond data science teams, is something that several companies have emphasized, including Google Cloud AI products like Kubeflow Pipelines and AI Hub as well as advice from the CI&T consultancy to ensure AI systems are actually utilized within a company.

The orginal article.

Summary of “Why Trying to Be Perfect Won’t Help You Achieve Your Goals”

One hundred photos would rate an A, ninety photos a B, eighty photos a C, and so on.
They would only need to produce one photo during the semester, but to get an A, it had to be a nearly perfect image.
During the semester, these students were busy taking photos, experimenting with composition and lighting, testing out various methods in the darkroom, and learning from their mistakes.
As Voltaire once wrote, “The best is the enemy of the good.”2.Start With Repetitions, Not Goals.
If you want to be a great photographer, you could go on a quest to take one perfect photo each day.
If you want to write a best-selling book, then you could spend 10 years trying to write one perfect book.
It’s not the quest to achieve one perfect goal that makes you better, it’s the skills you develop from doing a volume of work.
If you ignore the goals and build habits instead, the outcomes will be there anyway.

The orginal article.

Summary of “Don’t have 10,000 hours to learn something? All you need is 20 hours |”

Writer Josh Kaufman shares his own tried-and-tested technique to learn a new skill by putting in just 45 minutes a day for a month.
Wanting to learn something new comes from that best, most curious part of us.
Writer Josh Kaufman, author of The First 20 Hours: How to Learn Anything Fast and The Personal MBA: Master the Art of Business has figured out why so many of us get stopped in our tracks during this early learning period.
Through trial and error, he has come up with four steps that can help you scramble up the sharp slope of the learning curve in as little as 20 hours.
The first thing you need to do is to decide what you want to learn, and then break it down into smaller, manageable pieces.
Learn enough to know when you’re making a mistake.
“It could be books, it could be DVDs, it could be anything, but don’t use those as a way to procrastinate.” After all, you won’t learn how to bake bread or do yoga unless you break out the flour or yoga mat and do something.
To overcome what Kaufman calls the “Frustration barrier” – that period in the beginning when you’re painfully incompetent and you know it – you must commit to sticking with your new activity for at least 20 hours.

The orginal article.