simon-drouin-blog

[simon-drouin-blog]

One small step towards the necessary transition of human-computer interface towards physical interaction.

[simon-drouin-blog]

I would have like to attend this talk by computer graphics researcher Ken Perlin (like in Perlin Noise). From the abstract: "Sometime in the coming years — whether throughubiquitous projection, AR glasses, smart contact lenses, retinal implants or some technology as yet unknown — we will live in an eccescopic world, where everything we see around us will be augmented by computer graphics, including our own appearance. In a sense, we are just now starting to enter the Age of Computer Graphics."

[simon-drouin-blog]

Can't wait for being relieved of the burden of driving. Technically, we are there. At this point, it is more a question of social engineering than anything else. This project might be an answer to the problem. From a technical point of view though, this seems to be a clear case of what AI researchers call overfitting: learn to behave perfectly on a course that is just a small sample of what is out there and you end up with an algorithm that generalises poorly.

[simon-drouin-blog]

From the article:"In Eagleman’s essay “Brain Time,” published in the 2009 collection “What’s Next? Dispatches on the Future of Science,” he borrows a conceit from Italo Calvino’s “Invisible Cities.” The brain, he writes, is like Kublai Khan, the great Mongol emperor of the thirteenth century. It sits enthroned in its skull, “encased in darkness and silence,” at a lofty remove from brute reality. Messengers stream in from every corner of the sensory kingdom, bringing word of distant sights, sounds, and smells. Their reports arrive at different rates, often long out of date, yet the details are all stitched together into a seamless chronology. The difference is that Kublai Khan was piecing together the past. The brain is describing the present—processing reams of disjointed data on the fly, editing everything down to an instantaneous now. How does it manage it?"

[simon-drouin-blog]

Not the first work on the subject but something we will see more and more. In a world where it is increasingly difficult to spot digitally manipulated images, conspiracy theorist might suggest we are condemned to be manipulated by those who control images. But there is hope, researchers are coming up with ways to detect the fakes. It is probably the beginning of yet another race between those making fakes and those trying to uncover them.

[simon-drouin-blog]

By Dr. David Ragsdale, professor at the Montreal Neurological Institute.

[simon-drouin-blog]

This is certainly a big trend in the next few years: combining sparsely captured data with procedural methods to create highly detailed models of the world. This one is for trees, but a lot of other elements of a landscape can be modelled in a similar way. (via IEEE Xplore Abstract - Data-Driven Synthetic Modeling of Trees)

[simon-drouin-blog]

This paper aims to automatically improve the layout of a graphic design based on an optimization process that takes into account importance of the elements as specified by several users. Quality graphic design will probably always require human intervention, but it will become easier for non-professionals to come up with a descent result, just like the average joe today can take much better pictures than 20 years ago.

In the future, we should be able to create drafts for powerpoint presentations almost automatically. Users will dictate the presentation out loud and the software will automatically identify keywords and find illustrations from online image databases and layout everything automatically.