This how it all starts
Jan. 21st, 2012 07:44 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
ellestraThe NewScientist has a slew of robotics and AI articles. I look at some of these and I can't stop thinking this is how the whole things starts. After the world becomes a thing run by them these will be what they would look at as we look at our baby pictures - you know the in utero ones. Especially, The Painting Fool's art:
In another attempt to grow an AI are neural networks and this one taught itself to estimate the number of objects in an image without actually counting them. The concept of numbers were not explained to it but it learned to react to the number of objects and the pattern it's neurons create depending on number of objects is similar to the one observed in monkeys. So not only step forward the more efficient image recognition but also another proof of evolution.
Speaking of programs getting better at seeing a newly developed system is able to recognise faces even after plastic surgery. It uses facial-recognition technique similar to sparse representation, which compares individual features of a face with combinations of individual features from faces already recorded in a database, but can work with single image of before and after instead of multiple images from different angles. It can take to mugshots and tell if it's the same person. Not so long ago computers had problems with differentiating simplest images, soon they'll be better at it then us.
Another problem is how to make communicating back with humans more fluent. One way that can help is body language. Taking tips from dancing, masked theatre and cartoons to make robot act out their thought processes. Just like Wall-E.
This also a step toward making robots that are actually useful in everyday life instead of just something to display. Something we all could use in everyday life. How many years it's been since Roomba debuted? In the 21st century we were all to have robots in our home doing things for us. However robotics never really progressed as fast as processing and software development. Robots have still very limited use even in the manufacturing and military where they are most present. Time to get them to every home.
In another attempt to grow an AI are neural networks and this one taught itself to estimate the number of objects in an image without actually counting them. The concept of numbers were not explained to it but it learned to react to the number of objects and the pattern it's neurons create depending on number of objects is similar to the one observed in monkeys. So not only step forward the more efficient image recognition but also another proof of evolution.
Speaking of programs getting better at seeing a newly developed system is able to recognise faces even after plastic surgery. It uses facial-recognition technique similar to sparse representation, which compares individual features of a face with combinations of individual features from faces already recorded in a database, but can work with single image of before and after instead of multiple images from different angles. It can take to mugshots and tell if it's the same person. Not so long ago computers had problems with differentiating simplest images, soon they'll be better at it then us.
Another problem is how to make communicating back with humans more fluent. One way that can help is body language. Taking tips from dancing, masked theatre and cartoons to make robot act out their thought processes. Just like Wall-E.
This also a step toward making robots that are actually useful in everyday life instead of just something to display. Something we all could use in everyday life. How many years it's been since Roomba debuted? In the 21st century we were all to have robots in our home doing things for us. However robotics never really progressed as fast as processing and software development. Robots have still very limited use even in the manufacturing and military where they are most present. Time to get them to every home.
