On combining event processing with neural networks

This picture is taken from a recent article by Kamalkumar Mistry which explains the benefits of possible combining event processing with neural networks, starting by explaining each of these technologies separately and then providing a patient monitoring example. In this example the combination of the two technologies is that a patient is monitored in real-time, the signals from the monitors are getting to an event processing system as an input, and after processing is done the output is fed into a neural networks, which recommends action based on an  individual model of each individual patient.

This is a valid use case,  I actually thought about additional use of neural nets and this is to tune up the monitoring, in event processing terminology - set up the patterns to be monitored.   Taking back the monitoring patients example,  which is actually one of the use cases we analyzed in the past,  in fact it is example no.1 in the examples we put in the introductory chapter of the EPIA book, (chapter 1, page 7).  In our example, the physician can tune up the system to have individual monitoring patterns for each individual patient, since in different patients, different combinations of signals over time may mean different things.   The event processing pattern can be recommended by the neural network system based on the same patient model mentioned by Mistry, and in this case the relationship between the neural net and the event processing system are reversed.

Anyway - it is interesting to investigate this combination further.

A promo: stay tuned for the inauguration issue of the online magazine soon... 

On Dynamic EPAs by Berhnard Seeger

I came across a presentation by Berhnard Seeger entitled: "Dynamic complex event processing - not only the engine matters"  - the picture above is taken from that presentation.  Seeger uses the term "DEPA" for "Dynamic Event Processing Agent".   The dynamic refers to the ability to add/modify EPAs without affecting the event sources and event sources without affecting the EPAs, and ability to change EPAs at run-time (we haves supported this feature in Amit). 

The reference between all the players are indirect and done through meta-data entities.  There are other components to this model -- inclusion of actions in order to check contradictions, simulations for debug,

I totally agree that all these features are important (not sure that a new term is needed, this relates to implementation of EPA),  in fact we have worked on related issues in the past, see our paper in DEBS 2010 entitled: "analyzing the behavior event processing applications". 

In any event - interesting presentation, read and enjoy!   

The big data hype cycle 2012

I haven't written in the last few days,  I have been in EU project review (as a reviewer) in Brussels and also had some time to be tourist, and climbed the Atomium, Brussels known icon

and  in several museums in center city, taking refuge from the rain 

  including the famous Magritte museum.   I have imported some Belgian chocolate (most of it was already given away)  and a Belgian virus, with whom I am struggling in the last couple of days.

I also came across the Gartner's big data hype cycle for 2012 -- the first time in which Gartner chose to look at big data as an area.

You may notice that "complex event processing" is around the peak of the diagram.

It seems that this hype cycle made Irfan Khan, CTO of Sybase quite furious, his firm reaction was:

"Gartner dead wrong about Big Data life-cycle".    Khan claims that Big Data is not a hype but a reality, and expectations are under-inflated not over-inflated since it can do much more than what people assume.

I guess that there is growing adoption to technologies associated with Big Data, but I don't think that it reached the plateau of productivity, as Khan's claims,  since this is not around whether there are mature products (by the vendors' conception), but around the utilization in industry, and it is difficult to say that most organizations had good exploitation of such technologies.  Furthermore, Khan's claim that Big Data is under-inflated actually shows that the plateau of productivity has been reached.   

In any event,  the event processing angle is interesting.  Note that originally event processing appeared in the hype cycle of enterprise architecture for several years.  In 2012 event processing does not appear explicitly, 

Big Data appears as one block in the top.  This shows that event processing has migrated (at least in Gartner's mind) from the middleware world into the analytics world,  and this is also compatible  with some of the current trends, but this should be a subject of another posting - coming soon. 

On the "end of the engineer"

After writing yesterday on science and engineering somebody attracted my attention to a (not new) very visible  posting by Tom Gillis on the Forbes Blog entitled: "The *End* of the engineer".    Gillis, who labels himself as an engineer who grew up in a family of engineers claims that in the past what the market competition was on better engineering and brings some examples of high-tech vendors who failed due to the fact that others succeeded to get better engineering.    The claim is that it is no longer the case, the differentiation is not in the engineering, but in understanding customers needs (even if the customers are not aware of them),  the ultimate example is the direction that Steve Jobs took Apple whose success was due to the market insights and not to superior engineering.    While engineers are still needed, Gillis claim that now they are not the one who will bring the crucial value, but those who can understand the customer's way of thinking, thus the heroes of the high-tech will be those who have "soft" skills, and the education system has to reflect it -- interesting perspective,  as you can imagine, also controversial, you can view the comments to the original Blog posting, some of them had strong opinions to either side (the author added prefix to the Blog in response)...   Not sure it is the end of engineering -- but I agree that the education for high-tech workers today is not technology only...

On software, engineering and computer science

Today the IBM Haifa Research Lab hosted the Programming Languages and Software Engineering whole day seminar, the keynote speaker was David Parnas,  one of the pioneers of software engineering,  Many years ago when I was in the Israeli Air-Force, I have investigated the new discipline of software engineering to see whether we can apply in a big software project that I've managed, and Parnas was one of the first names I have encountered, but this was the first time I saw him in person.  There was a panel on the term and profession of software engineering.   It reminded me one of the posts on this Blog from early this year that was entitled "Is Computer Science - science or engineering?" which was triggered by the fact that my daughter Daphna had a "science day" introducing the science classes in the high school she started to attend this year, and despite the fact that they teach "computer science" they did not include it in the science day - thus the school does not think it is a science.    

Interestingly, David Parnas, as well as other panelists, don't really think of computer science as engineering, in fact, David Parnas talked about a fight between an engineers association in Canada to forbid computer science graduates that don't really have engineering training to call themselves engineers.   I always thought that there is something pretentious in the fact that programmers (typically with computer science education) call themselves "software ENGINEERS".  As somebody said in the discussion today  -- an engineer is a person that can be sued on negligence if it will be proven that engineering rigorous principles were not met.

Until the time it will be true for software, software producers cannot call themselves software engineers.

Back to - is computer science a science -  Winton Cerf wrote in CACM an article entitled "where is the science in computer science?"   The answer according to Cerf is  that unlike physical sciences which are about modeling the world,  in computer science the science is tools for understanding complex software systems and make predictions about their behavior --  not really sure I am convinced...

It seems that computer science might be an animal of its own - neither science nor engineering, especially in the conventional terms.

On the Internet of EVERYTHING

Cisco came out recently with the concept of  "The Internet of EVERYTHING".

While the "Internet of Things" deal with connecting anything to the Internet, the Internet of EVERYTHING deal with the things and the semantic connection among things that can make the world actionable in real-time.  A simple example is the car theft example. 

A car is connected to the Internet through its GPS sensor that reports its location, it has also semantic relations to a list of eligible drivers who are permitted to drive in this car, each of these persons is also connected to the Internet using his or her mobile phone, thus the Internet knows the person's location  (disregard the privacy issue for this scenario!),  so if the car is moving (inferred from the GPS change in location), and all the eligible drivers are not in the car - it means that the car is stolen, and it can then report to the police and have them track its location.  

This scenario is based on - things,  contexts of things, and processing events about things..   It is actually quite straightforward from technology point of view.  I wonder if the "Internet of EVERYTHING" will survive the buzzword test of time... 

On IBM scientific accomplishment

This week, the annual accomplishment process of the IBM Research Division was concluded. This is a process that recognizes major impact activities in various categories: scientific, contribution to IBM products, contribution to IBM services, contribution to standards and some more.  

Within this year's process, our work on the event processing conceptual model has been recognized as a scientific accomplishments.  The criteria are: number of citations (according to Google Scholar) and support letter from senior members of the scientific community in this specific area. 

It is interesting to note that the major publication referred was the book I have written together with Peter Niblett, "Event Processing in Action".

 The interesting fact is that the book was not written as a research oriented book, but was geared towards the professional market,  yet it accumulated so far 153 citations, with the number steadily growing (when the process started the number was around 130).  

Drilling down to the citations list it is also interesting to observe that while some of the citing papers belong to the event processing community, many others come from different domains and implemented systems in the areas of power management in mobile devices from Finland, rotor-craft control from Brazil, as well as others that indicate that the material in the book had some practical impact in additional to the impact on the scientific community, which is also important, as science is being built in layers.

I have been out of the research work for about 10 years, where I kept research activity in the back sit, mainly through supervising  PhD and MSc students at the Technion.  The major project I was involved in the years 1998-2005 (AMIT), has a single major publication that is actually the summary of the PhD dissertation of Asaf Adi  (this paper also accumulated nice number of citations). 

The question whether citation number is a good metric - is another discussion, for me the actual impact (those using the work in practice) is also an encourging indication that the work is not done in vain -- more later