Hey everyone, big news this week! Apart from your usual injection of concentrated AI-news, we also welcome our first guest: meet Antje Farnier, astroparticle physics researcher turned deep-learning expert. Make sure to subscribe to the blog if you don’t want to miss the next one!
1- Put your seatbelt on!
Who hasn’t dreamed of jumping into a fully autonomous car just like Tom Cruise in Minority Report? If you read recent headlines, you might think that a self-driving car is just around the corner. Unfortunately, it’s not. And we might need more time than we thought before it becomes a reality. Even though technologies made a huge leap in real-time object detection, it’s not sufficient yet. Whether you agree, disagree or want more details, you should definitely have a look at Michael Guo’s latest review: Self Driving Cars, The Most Hyped Thing Since… The Segway?
2- Get your hands dirty
Deep-learning is no different than anything else: if you want to be good at it, then PRACTICE, PRACTICE, PRACTICE! What better way to practice than to start your own little project? For instance, you might want to develop your own in-house AI-assistant, just like Zuckerberg did last year. Soon enough, you’ll realize that you need plenty of other skills such as electronics or low-level programming to name a few. The sad truth is: sometimes it becomes too overwhelming and all you want to do is to drop your side project. But it doesn’t need to be like this. Have a look at the AIY Projects and start with the Voice Kit. You’ll be rolling in no time!
3- Enough is never enough
Companies are realizing that their internal data might represent life or death in tomorrow’s AI world. They’re putting it in vaults first, before trying to find out what they can do with it.The new gold, that’s how valuable data has become. Whoever you are — academic, startup wannabe, big tech company — chances are, if you’re using any kind of AI/Deep Learning algorithm they run on data! And as with any car, if you want to go further than your competition you have two choices: get better mileage or simply get a bigger gas tank. The thing is, data is not easy to come by, and even then, you have to fight for it. That’s when open-source datasets come into the picture; they’re often the first step for testing an idea, but their number is limited. Well, my friend, count yourself lucky, because three brand new datasets just got out! Have a look at the MNIST of fashion, all the bad drawings you can think of, or Alphabet’s latest release of speech commands!
4- Now what?
Little Ripper protecting the beach so that you only have to think about sunscreen.You’ve made it this far. You have data to play with and the skills to extract information from it. And then it hits you: what should I do with it? We sometimes fall into the trap of gathering a lot of data without thinking first of what we want to do with it, which can lead to some lesser-extent existential crisis as we realize that it’s only the first part of the journey. If you’re stuck here, it’s sometimes helpful to take a step back and look around for inspiration. This week, inspiration takes the form of a shark-detecting drone named Little Ripper patrolling the Australian coast, Alibaba’s new feature that let’s you pay with a smile, and if you’re more business inclined, how the ad-world is being disrupted by AI.
5- Get your headphones ready!
It’s not always easy to keep up with the daily news and breakthroughs of the AI world. There’s so many that it becomes difficult to know what’s worthwhile your time. Plus sometimes you don’t have the time, or simply can’t read your favorite newsletter. If that’s the case, you should then definitely check out Courtney’s top 5 AI podcasts. In future, while driving your car, taking the subway or going for a run, you’ll always be able to get your AI fix!
