Applying ML to IoT, Bad Design Kills, and My Facebook Data Selfie — The Weekly Roundup

Check out the latest posts from IoT For All! Click the titles or pictures to read the full stories.

1) Applying Machine Learning to the Internet of Things

The data models that are typical of traditional data analytics are often static and of limited use in addressing fast-changing and unstructured data. When it comes to IoT, it’s often necessary to identify correlations between dozens of sensor inputs and external factors that are rapidly producing millions of data points.

While traditional data analysis would need a model built on past data and expert opinion to establish a relationship between the variables, machine learning starts with the outcome variables (e.g. saving energy) and then automatically looks for predictor variables and their interactions.

In general, machine learning is valuable when you know what you want but you don’t know the important input variables to make that decision. So you give the machine learning algorithm the goal(s) and then it “learns” from the data which factors are important in achieving that goal…

Calum McClelland explains the applications and use cases of Machine Learning in the Internet of Things.

2) March Madness or Hoop Genius? You Make the Call.

I hang out with several die-hard college hoop fans and former players that have an extra bounce in their collective steps this time of year. For most of them, it doesn’t get any better than March Madness — even the Super Bowl pales in comparison.

That got me thinking, could IoT technologies help transform most people’s mediocre games into a “nothing but net” experience?

You’ve probably seen or heard of the smart basketballs that come with a companion app that connects to your phone via Bluetooth — tracking shooting accuracy, range, percentage, and other statistics. They’ve been around for several years now so I wanted to check up on their progress and see if the hype was real.

Here’s what I found out…

If you’re also curious where smart basketball technology stands, Eric Conn will fill you in!

3) Bad Design Kills: Self-Driving Cars or Not

For 30 years, it was assumed that using crash dummies modeled after a 6ft tall male would adequately test for the safety of both men and women. And before crash dummies, we used live animals and dead people. Don’t believe me? Spain still used them as of 2013 and when crash dummies first came out in the 80s, they were so popular they had their own toy line.

Overall, women are killed in crashes at disproportionally higher rates than males, because they were never included in safety analysis.

Females make up one-quarter of all driver fatalities and one-half of all passenger fatalities, but men drive 50% more than women and average 5,000 more miles driven per year. It wasn’t until 2011 that the federal government changed out the average male dummy for a smaller female dummy in some tests (even then, this didn’t include any tests in the driver’s seat). And even then, the single female crash dummy they use is 4'11 and 70kg. Not exactly your average woman…

Self-driving cars might be just around the corner, but Hannah White shows us why designing for safety needs to be the priority.

4) IoT Is Hard

With the Spreadsheets, PowerPoint’s and Reports behind you, now for the bad news — the software that is necessary to reach from the sensors to the enterprise is the integration glue and requires deep connected systems experience (skill sets intimate with embedded systems, connectivity, and real time, time series based systems) to develop industrial grade, end-to-end systems.

These capabilities are not part of your traditional corporate IT DNA, so if your assumption was that you were going to license an IoT platform, provide your enterprises developers a little training and access to the cloud, think again.

The challenge is further exacerbated given the lack of industry standards, a battle the titans will surely drag out, creating confusion, complexity, risk and lock-in scenarios. From a production ready perspective, I submit to you, one of the significant drivers behind IoT’s delayed traction is the supply-demand imbalance for these constrained connected systems competencies…

If IoT were easy, everyone would be doing it. David Houghton compares the current state of the IoT industry to the shift from centralized computing to distributed computing in the late 80s then shares his counsel for getting started in IoT.

5) My Facebook Data Selfie

A couple weeks back, I came across a Chrome extension called Data Selfie that collects data from your Facebook newsfeed and applies machine learning to give a snapshot of what Facebook might know about you through your digital footprint.

According to the creators, Hang Do Thi Duc and Regina Flores Mir, they aim to explain how our “data profiles, the ones we actively create, compare to the profiles made by the machines at Facebook, Google, and Co. — the profiles we never get to see, but unconsciously create”.

Although I only post on Facebook once every couple months to let the world know I’m alive on social media, I was nevertheless curious as to see what Facebook would conclude about me. The setup process was really easy. Just download the Chrome extension and let it collect data on the background while you browse…

Yitaek Hwang wanted to find out what Facebook knows about him while he scrolls through his newsfeed. Apparently he’s .06 negative on Silicon Valley and 0.57 positive on global health, among other things.

IoT For All is brought to you by the curious engineers at Leverege. If you liked this week’s roundup, please recommend or share with someone you think would enjoy it! Thank You!