On the 8th of July 2017, I attended UBB Graduation Days, a public event where local IT companies get the chance to award prizes for graduates' effort. I don't know how many people are familiar with this event, but I can say that of all the events I attended over the last years, it had the largest dose of ingenuity and passion for technology. This event represents BBU Computer Science students' work for their BA thesis. Besides enthusiasm and dedication, the demos reflect present-day technologies very well. Many of them featured at least a machine learning component, combining various learning algorithms in order to solve a specific problem.
Since theory is more enjoyable if presented under the guise of a practical example, I will briefly present some of the projects exhibited:
A small robot, Arduino, which detects the walls in a chamber through ultrasonic sensors, determines their height and generates a 3D map of the environment
A machine learning application that identifies commercials in photo streams like Instagram, or in user-sent photos. Further developments will be able to signal misuse (a vase with an Adidas logo, for example) or suggest a greater-impact positioning of commercials on a webpage
Augmented reality was present in a project made of an Arduino client and a server. The phone recognized the surface recorded by the video camera and projected a sphere composed of six loxodromes, sent by the server. I noticed the quality of the graphical representation, the good recognition of the surphace on which the 3D image was projected. Of course, the system's performance could be improved, but generally it's a project written with a lot of passion and which is well implemented.
An IOT application that controls sensors on a remote board. The demo consisted in lighting a LED when the temperature sensor exceeded a threshold (recording from a new sensor on a certain pin)
Breast cancer detection using machine learning. Through a rather limited data set, ML algorithms were trained for its recognition. Moreover, there was a separate application that focused only on the tumor itself.
An AI-based framework employing two practical image recognition algorithms. Unfortunately, I could only see a PowerPoint presentation, not a demo, since the UI part is not developed yet.
A web platform based on machine learning which detects musical genres. The demo was impressive: in a short time, it catalogued Queen's Bohemian Rhapsody as a rock song. The recognition strategy is noteworthy: several agents were used, for each genre: rock, jazz, classical, electro, metal. At the end, each of them publishes the recognition rate for a genre, and the best one wins.
Everyone I spoke to was very pleased with their applications and wanted to continue them. I also noticed the presence of many local IT companies who were there to reward Innovation, either through prize money or the possibility of receiving technical support for one year in order to grow the startup. I have to remark that all those who I spoke to were already employed, and unfortunately none of them could pursue the chosen direction and technology at the workplace. I would have liked to hear at least one of the graduates that they were trying to launch a startup on their own, and that they don't want to get employed for the time being. Certainly, a deeper involvement in the enhancement of these products designed by the young graduates would narrow the gap between the outsourcing and local innovative product market. At the end, I would like to congratulate associate professor Simona Motogna for organizing this event and for her effort in promoting the collaboration between the academic environment and the private sector. The event was a success and we look forward to attending the event in 2018 too!
by Dan Dragu
by Mihai Varga , Raul Deac
