Been working on a very interesting initiative in space of Machine Learning analytics but now trying to get back in groove of things and continue exploring some of the emerging technologies, ask hard question, and see how we (as a community of technologies) see the world evolving as we going through all kinds of transformations that are pushed both by business initiatives as well as technologies.
Let’s begin our conversation on the topic of time warps and self-learning behavioral models.
Here is how I would like to solicit the feedback on this (in my humble opinion) interesting problem. All the comments please submit through Quora which but I will explain it here in more details.
Let’s say you creating a “robot” that is responsible for possibly some mission critical operations, example of such could be cloud operations management where failure could lead to a loss of a customer of a lawsuit or maybe even some healthcare related operations where failure could lead to as unfortunate consequences as death. Now, the “robot” leveraging some learning principals to be able to obtain the definition of norm that is captured in variety of numeric and inner-playing data features (IT stats or human body stats). The definition of norm is derived from learned behaviors that (as you can imagine) are very much time specific (the way IT/cloud is being used or the way human behaves).
How do you program a “robot” that it is capable to deal with daylight savings time? I have a few solutions in mind, and NO UTC is not necessarily an option as any computer scientist would approach (see background and derive it from a hint that we are dealing with behaviors and go from there), but would like to hear from the community. Perhaps it is use case specific, but still sharing some ideas here would be interesting. Looking forward to it!
Link for thoughts and comments: https://www.quora.com/How-do-you-deal-with-daylight-saving-in-self-learning-behavioral-models