Tool to determine how easy a photo is to remember

This is an online tool based on the artificial neural neural network developed by the team at MIT with the ability to determine exactly which part of the image is easy to remember or remember. The goal of the study is to learn about human memory activity but they also "widely" release this tool for us to try.

Memory is a miraculous gift that creator has given people that if not, we will not be able to learn, cannot learn from mistakes, cannot identify friends and travel around this world. Scientists still do not understand what makes a particular problem easy to remember or forget and to do this, researchers at Massachusetts Institute of Technology (MIT) have developed a technique. image scanning math.

After conducting many universal studies, the team created a twisted neural network (CNN) with the ability to accurately predict the "memory level" of an image. Basically, CNN is an artificial neural network designed to simulate the arrangement of menstrual cells in the cortex in charge of visual organs, processing image information. These networks can deep-learning themselves, processing a large amount of data to identify hidden patterns in them. In other words, they can self-learn information without requiring advance programming.

The process of analyzing the memorization level of a photo with scores on a scale of 0 to 1.

In the report published by MIT's Computer Science and Artificial Intelligence Laboratory, the team explained the process of implementing a series of mundane experiments that define human memory. In it there was an experiment that showed a sequence of images, some repeated, and asked the volunteers to press a button each time they discovered a photo that had been seen earlier.

Next, the researchers analyzed the images to determine the characteristics of an image to determine what is responsible for helping us remember. Typically, the group found that it is easier for people to remember photos than for scenes. And summing up the research, the team created an algorithm that can predict the ease of memorization or forgetfulness of a photo.

Accordingly, CNN can achieve a correlation coefficient of 0.64 . With this score quite close to the ability to identify human objects, it shows that this algorithm is fully capable of predicting the level of visual memory. The group calls this algorithm MemNet and according to them, the team thinks that the algorithm can be applied very broadly into practice. For example, understanding the nature of memorization, one can enhance those elements to ensure that viewers can remember more easily and deeply.