Recap: Getting to Grips with Algorithms! (cue mighty roar)
As soon as I walked up the stairs in the TinEye offices on Queen Street East, it smelled overwhelmingly of grilled cheese. Right then, I knew Getting to Grips with Algorithms! (cue mighty roar) was going to be a great event.
Not only did the TinEye crew deliver a fantastic spread of snacks (gourmet grilled cheese, bowls of candy, and mason jars of homemade custard with berries and veggies & dip) speakers Leila Boujnane and Inmar Givoni captivated the crowd with their fascinating, funny and relatable talks. The goal was to leave the Girl Geeks audience with a better understanding of how algorithms, a concept that Google, Apple and Facebook have brought into our mainstream tech vocabulary, are changing the world around us.
Inmar is a member of Technical Staff at the Toronto Technology Centre office of Altera, and started us off with an intro to Machine Learning Algorithms (her specialty). Put simply, algorithms translate things out in the world into numbers, then crunch those numbers. They easily solve tasks that are annoying and time consuming for us humans, like looking up someone’s name in a database of a million names. Algorithms have migrated away from strictly academic uses into more mainstream areas where we all enjoy their analytical power on a daily basis, like the voice recognition software of iPhone’s Siri.
But there are still some tasks that are difficult for computers, like visual processing (Inmar’s example: a video clip of Star Trek heartthrob Ryker, the evening’s mascot). While the human cortex processes visual information easily and intuitively, computers can have a much harder time with it.
How do we write software tools that solve these problems? Enter the Machine learning approach. This means gathering lots of data, allowing a computer to process it, and letting the computer write the data-sorting rules for itself. A lot of progress in the accuracy of machine learning algorithms can be attributed to the large amount of data available to us today (more good quality data = better results). But we still have a long way to go if we want computers to be able to capture more nuanced visual information the same way that the human brain can; for example, the nuances in an having to do with social context, subtle facial expressions or humour.
Leila Boujnane’s half of the evening capitalized on her expertise in image recognition. Leila is the Co-Founder and CEO of Idée Inc., creators of the reverse image search engine TinEye, a great example of how image recognition algorithms are being used in the Toronto tech industry right now.
Leila highlighted that the world of algorithm and analysis is also changing because of the reduction of the cost of and the increased in processing capability of computer hardware. She wowed us with some great examples that illustrated just how accessible image recognition has become today. For example, it’s used in the facial detection features in most new digital cameras, in automated checkouts in grocery stores (detecting the shape of the objects in your cart), in modern car safety features (in Europe many new cars have pedestrian detection systems), as well as in more scientific tools like human iris comparison (which allows scientists to match the unique irises of two individuals to confirm identity). In a suggestion fitting for a Toronto crowd, one audience member marveled at how useful this technology would be if cars could detect the bicycles riding next to them in our streets, preventing collisions.
In short, Leila showed us that image recognition is already changing how we live and will soon be embedded in everything, all thanks to these sophisticated algorithms. If you missed the event, but want to get a better idea of image recognition, I suggest trying the publicly available and free version of TinEye.
For more info on Inmar or Leila, check out their bios here.