The Russian job search service Superjob has developed a neural network that allows, using a special algorithm, to fill out the resume of an applicant for a position using his photo. Despite the lack of other data, this summary is 88% accurate.
“A neural network can already easily determine whether a person belongs to one of 500 basic professions. For example, with a 99% probability, the system will distinguish a photograph of a driver from an accountant or a salesperson from an environmental engineer,” Superjob told TASS.
Also, with a 98% probability, the algorithm allows you to determine gender, age, work experience, and higher education. Moreover, with its help you can find out what salary the applicant expects.
The algorithm was calculated based on the analysis of 25 million photos from resumes. Its developers also created a clothing database of more than 10 million samples. “We know how much these clothes cost. After all, the saying “You meet people by their clothes” didn’t just appear. Therefore, depending on what a person is wearing... the system will calculate the salary appetite of the applicant,” says Alexey Zakharov, president of the service.
The developers noted that the more the applicant’s photo matches his profession, the easier it is to create a resume. After this, the applicant will have the opportunity to correct it independently.
Source: 3dnews.ru