On April 22 and 23, the Machine Learning Prague conference, with around 500 attendees, took place in Prague. Our IT Hub was represented not only because of our colleagues who attended it, but we also had our stands there as well. At the stand, we provided participants with details about our company as well as some projects connected with Machine Learning. Each day, we had only three breaks during which we could attract attendees´ attention – and we managed it! We ran an IT/Science-related quiz where more than 150 people participated. It was a great success for us, because it was more than 25% of the approximately 500 participants.
Below are some highlights from our colleagues who attended the conference:
Andrej Palicka shared:
“Most of the talks on the conference sparked my interest; however, there were two talks that I particularly enjoyed.
First was the talk from Chris Wiggins, the Chief Data Scientist at The New York Times. He described how advertisement giants such as Google and Facebook affected the ad revenue of publishing houses and how he and his team, as data scientists, need to tackle that. One of their products that I found interesting is a service that recommends which articles should be published to which social network at the right moment to gain maximum viral growth. Another one was the analysis for the supply chain of the newspapers, where they built a model that recommended how many newspapers they should deliver to individual sales points.
Another highlight for me was “Serving a Billion Personalized News Feeds” by Lars Backstrom from Facebook. It brought some insight into the kind of machine learning problems Facebook engineers need to tackle. To make the use of Facebook enjoyable for everyone, they need to personalise each news feed, so that the users see things they care about. For this, they need to figure out a person’s interests, key people they want to know about, their age, etc. And, all this on a huge scale.”
Jakub Smid shared:
“Chris Wiggins from The New York Times raised important questions about protecting democracy. As the revenue of the company declined from 3.4 billion to 1.5 billion USD, The New York Times faces one of the greatest challenges in its history. The lost revenue went directly to Google and Facebook. The difference is that these tech giants have no incentive on keeping an eye on democracy as The New York Times with its teams of investigative journalists. This transfer of readers is alarming, especially in an era of fake or misleading news.
He also answered a question about whether The New York Times will move to computer generated journalism. The answer is ‘no’ as they do not want to become yet another company generating reports. They want to keep their brand as one of the best newspapers in the world.
I was also impressed with the talk by Martin Vejmelka from Avast. He showed the details of their machine learning pipeline that clusters files based on their level of infection. You can explore the clusters in Star Wars like virtual reality, quickly see the difference between the files of the same cluster, and find the feature that could distinguish the harmless files from the malicious ones.”
Jan Tkacik shared:
“For me the most interesting part of conference was Pierre Baldi`s talk about using machine learning in life sciences. He showed many fascinating examples from chemistry and medicine which can be inspiration for our future projects but also pointed out many obstacles to overcome.“