Monday, December 1, 2014

CA welcomes much needed rain

Finally the rain is here in California!
California has seen the worst drought in history (ask @JayFamiglietti, for example!).

I would be interested to see how the map is updated after these rainy couple of days:
http://droughtmonitor.unl.edu/MapsAndData/WeeklyComparison.aspx

Howver, this weekend and through few of the weekdays, we will see some rain. Below I have copied snapshots of weather by NOAA.



The situation is almost like a welcome a song from the Bollywood movie "Lagaan" where they sing to the incoming rain.



Watch the beautiful GOES west view here:
http://www.goes.noaa.gov/goes-w.html
---
-->

Friday, November 21, 2014

How to make interacting robots loveable: allow them to make small harmless mistakes

The other day I was chatting on a chat service to solve some issue. These days the websites offer "instant help" with live agents. Chat bots have become very smart and progress in machine learning for symantics and ontological reasoning has progressed by leaps and bounds. An example: cleverbots
http://www.cleverbot.com/

Anyways, to begin with, I was greeted by a chat bot, which tried best to mimic humanly possible replies: with empathy and dutyfully. What I found to be cute was a typo it was making.
I think allowing the robots make small--harmless mistakes: such as a typo in this case, would make them more personable.

We expect the machines and robots to perform flawlessly. In critical sector that can be highly important. You want to avoid mistakes at 100% and be precise 100% of the time. But, think of interactive machines... so that people are not bored, it would make sense. Don't you love the sound of child making minor mistakes and going "Oh, oo!".  That kind of mistakes.

However, it has to be within the limit of being annoyed though. For example, the first mistake in the chat was OK. The second becomes kind of weird.



---
-->

Monday, November 10, 2014

Intelligent Wifi Channel Selection

My apologies up front if you landed here (thanks G) seeking for an algorithm for automated wifi channel selection. This is a very basic rant.
/rant-on
The internet router has a setting to avoid overcrowded channels. By setting it to decide on the auto channel, it is supposed to choose the one which is the least crowded. The problem is that there are other routers on the air doing the same thing. The end result: they end up in the same channel. (Well, I think they all are of the same kind.)
Here is a photo proof:
16 of them selected to be at 11, probably because all of them at some point figured that 11 was the best one!

Clearly, the plan is self defeating. A general advise on channel selection (word of mouth) is that channel 5 and 10 are usually crowded: "avoid them!!"
Well, if I am not wrong, after repeated tests (figures not included), I can conclude that the above statement is wrong!

Therefore, I have decided that it is not optimal anymore to let it auto select the channel. Because according to the game theory, they would play the same (defeating) game. The result is shown above where they love channel 1 and 11, which are supposed to be least attractive.  By going into a set channel, say 10, or 2 or 5, or 8, whatever, you know what you are going for.

Have you set your router to an intelligent wifi channel selection?

/rant-off
---
-->