Computers with Senses

Computers from what they have been in the past have been by design and purpose, faster and more efficient than our biological selves. Sometimes, we have used our superior attributes to improve them and sometimes we have used theirs to improve our own. The biological body, whether human or otherwise, is complex and much of its reasons and meanings are still relatively unknown. For example, we know which part of the brain does what. What we don't know is how it does that. 



But perhaps, that day is not far off either, as some researchers around the world are now proving. Some interesting research in the past couple of months has also challenged the pessimist premise that computers are here to replace humanity, not aid it. Work, like the much-publicized foray of a software giant into the field of medical research seems to reinforce that comfort factor.

I know who you are!



The way we think and understand our own sensory systems applied to the way computers work, seeks to revolutionize and re-write our future. As humans, we have five senses-sight, touch, sound, taste and smell. Essentially, these senses send digital information to our brains. A computer is a digital device and it can use its own sensory systems-camera, microphone and so forth-in much the same way. Inputs are not limited to what we traditionally find on a desktop, but could include even applications such as radar, sonar, pressure and temperature sensors. Predefined hierarchies would receive this streamed information and then process it accordingly. Jeff Hawkins, the inventor of the Palm Pilot, has come up with this brilliant idea and recently demonstrated at the PC Forum 2005

(www.pcforum.com) that his algorithmic software can do this. In humans, it would be called 'associative memory', and we would equate it to artificial intelligence when computers do it. 

How it works



His systems works by recognizing patterns and then making predictions (intelligent guesses) based on that. Over a period of time, the system gains familiarity with its surroundings, similar to a baby. At present, this is limited to recognizing something that was previously shown to the computer and its modified counterparts. For example, you would first show it a picture of yourself and allow it to store the picture. Then you could show it other pictures of yourself-wearing different clothes, different hairstyles or in different surroundings-and each time, it would identify you even if you were facing the other direction! Currently, Hawkins' software can recognize up to 90 black and white images, even if their sizes are changed, flipped or rotated or distorted. 

At the lowest level of the sensory hierarchy, the input data would appear as mere fragments and measurements of various kinds. But, the right software algorithms and context could recognize this information at progressively higher levels of the hierarchy as understandable patterns. It is expected that such research would let us understand problems in new ways. The applications of Hawkins' work could range from weather analysis and forecast to tracking the stock markets. Such gray-matter driven systems would some day perhaps find a cure for life-threatening diseases such as cancer and AIDS.

Better recognition systems



Realistically, we could find such applications for security where they could solve problems with modern biometrics-based security. Modern finger-print readers for instance, would fail if you had a cut or a bruise on your finger- the reader will not be able to compensate for such a discrepancy and eliminate it as a false result. 



The Jeff Hawkins based engine would recignize it as the same finger. Why only fingers! The same system can be applied to iris-scanners. Authorities at sensitive guard-posts like at airports, would use such systems to identify terrorists and arsonists faster, since the system uses almost human-like recognition techniques to analyse images, instead of comparing pixels and forming approximations. This way, the program will not be affected by added scars, changed facial shapes and hair patterns like conventional systems would.

Microsoft patch for HIV vulnerabilities 2006



Systems applicable: humans with HIV infection. Don't panic, that's not a real product. Yet. The software giant's research wing (Microsoft Research) is working with doctors and scientists to find medicines to fight the spread of HIV. Their research is aimed at both combating prevalent infections as well as finding a permanent vaccine. Very strange, guess what the logic is behind such a research. Spam e-mail! 

Apparently the algorithms that power modern anti-spam engines could be re-written to seek out biological viruses that are often polymorphic and might exhibit similar properties to conceal themselves from anti-agents and to spread

itself. Some of the other technological areas being studied for porting over to HIV research are: file

compression and database management.

News reports suggest that first-phase vaccines are already out there, undergoing trials. We hope that when they're done ridding the world of HIV, they might start on the common cold virus and finding a cure for various Hepatitis strains. And hopefully, humans won't

need to reboot after getting the vaccination.

Putting the two together



Perhaps Jeff Hawkins' engine could be married to Microsoft's engine. This could chop the analysis time by a lot, resulting in better progress. If you watch the promo-video on Microsoft's website

(http://research.microsoft.com/displayArticle.aspx?id=1238), you will

notice the scientists complaining about not having a system that could 'see' patterns and analyze them quicker. Well, Jeff's engine apears to be exactly such a thing.

Taking it from there, the possibilities appear endless. Ranging from faster solutions to all of Human problems to maybe inter-stellar exploration and travel.

Sujay V Sarma