CategoryInvestigación @es

Página Web del INFER Lab

Como tal vez ya hayan leído, mi grupo de investigación tiene ahora un nombre y una página web. Soluciones GBH nos ayudó con ese proyecto, y creo que tenemos ahora una plataforma muy eficiente para comunicar quiénes somos y qué estamos haciendo. Si no recibieron la noticia, vengan a visitarnos:


Sobre lo que tal vez no hayan leído, es los detalles de la plataforma. Así que si están interesados en algunos detalles un poco más técnicos, aquí está un poco de información al respecto:

La web completa está usando WordPress Multisite, que era un proyecto separado de WordpPress y que recientemente fue integrado al proyecto WordPress (WP) principal. Tenemos custom post types para los diferentes elementos de la página (publicaciones, proyectos, investigadores, etc.) y taxonomías para ayudarnos a organizar y relacionar las entradas. De ahí que agregar un nuevo paper es tan fácil como crear una entrada en un blog, y todos los enlaces a proyectos relacionados o estudiantes, se pueden hacer desde la misma interfaz.

Para uso interno, cada uno de nosotros tiene un sitio WP bajo el multisite, el cual utilizamos como un cuaderno de apuntes para compartir hallazgos o problemas de investigación con los demás miembros del equipo. Cualquier entrada en ese sitio WP puede ser publicada en el sitio web principal directamente, si es que se quiere compartir con el público en general. También tenemos un sitio más, particularmente diseñado para las colaboraciones más casuales. Para ése, usamos el muy útil P2 theme. Hasta ahora, eso se siente como tener nuestro propia plataforma de Twitter interna, sin el límite de 140 caracteres.

Algunos plugins para WP que hemos estado experimentando son: bib3html, SyntaxHighlighter and MathJax-LaTeX.

Y eso es todo. Espero que nos visiten a menudo para saber en qué estamos.


During the last few weeks, with all the travels, conferences and exposure to other people, I have been ruminating a couple of ideas about the guiding principles that I think should be followed when conducting engineering research. Perhaps they apply to all types of research, or even to life in general. I don’t know. I do know that I would like them to guide my work.

As with almost any profession in life, in research it is easy to replace or confuse self-perceived success with success as perceived by others.  In other words, sometimes we loose sight of what initially drove us to this profession and instead focus on ensuring we are seen as successful by our peers. This is perhaps the observation that got me thinking during these past few weeks. I have clearly fallen for this trap and wanted to find a way to prevent this from happening.

The thoughts have been piling up and, to make sense of what is going on in my head, I decided to spend a few minutes (now an hour) writing down a short list of these guiding principles; a credo of sorts. It is still incomplete, but who knows if it will ever be complete. So without further ado, here is the result:
Continue reading

Two Calls for Papers on Energy Efficiency in Buildings

As if it were not enough, besides SustKDD 2012 and the 1st NILM Workshop this year I will be helping to organize another two events that could be of interest to those of you who are working on energy efficiency in buildings. Here are the details:

BuildSys 2012

ACM BuildSys 2012

To be held at Toronto, Canada

This has become a very interesting venue to bring together researchers from different areas. Although the main focus is still on the use of embedded systems for energy efficiency, in its fourth version the workshop is trying to expand the number of submissions from civil engineering, architecture, human-computer interaction, etc.

Important Dates:

  • Paper submission deadline: 07/30/2012
  • Workshop Date: 11/06/2012

Special Issue on ASCE’s Journal of Computing in Civil Engineering

Co-guest editing with John Taylor from Virginia Tech.

Co-guest editing with John Taylor from Virginia Tech.

The American Society of Civil Engineer’s Journal of computing in Civil Engineering is devoting a special issue to Computational Approaches to Understand and Reduce Energy Consumption in the Built Environment. The purpose of this special issue is to collect and publish a representative set of manuscripts from AEC/FM (Architecture, Engineering, Construction and Facility Management) researchers working in this field.

Important Dates:

  • Proposals due by: 07/31/2012
  • Full paper due by: 11/01/2012

For more information, follow these links:

Summary of the 2012 NILM Workshop

NILM Workshop Picture

2012 NILM Workshop at CMU

Last Monday, May 7th, the 1st International Non-Intrusive Load Monitoring Workshop took place on campus. Zico Kolter and I had been organizing this event for the past few months, in an attempt to bring together the community of researchers and industry practitioners who are working on electricity disaggregation. By all measures, the resulting event exceeded our expectations. We had a great turnout (60 participants) and a very interesting lineup of speakers and talks.

The paragraphs below are a summary of what happened during the event, so if you attended or are simply interested in getting access to the slides and papers here are some useful links:

  • Most of the extended abstracts have been posted online, along with posters and slide decks for the presentations.
  • We have also published a few photos of the event here.

Continue reading

EPRI Load Shape Library & Customer Load Insights Interest Group Workshop

The semester has still not come to an end and yet the summer travel season has apparently begun. I have been in Dallas since yesterday, attending a workshop on load shape libraries and customer load insights arranged by the Electric Power Research Institute (EPRI). Tomorrow (more like in a few hours) I leave to New York to participate in a panel as part of BuildingsNY. Then there is a trip to the Wolf Creek Dam in a few weeks, and more trips after. All in all, I will only be in Pittsburgh for about 14 days this month. OK, enough about my travel schedule.

Today during the workshop I jotted down a few simple and unprocessed ideas/observations that I would like to share. It is always interesting to participate in conversations with industry because the disconnect between academic research and industry problems becomes evident very quickly. Most of these disconnects are due to the fact that the answers being sought by academia will only become useful later, but often times these answers are in response to problems that do not even exist and will likely never exist; and that is a big problem. The observations I made today are not necessarily under this category, however:

  • There is a need to quantify the benefits of load research. Although there are no hard numbers, the benefits of customer research are believed to be:
    • The possibility of making informed decisions to defer large capital investments.
    • Informing policy and regulation with solid evidence.
    • Billing dispute resolution
    • Fraud detection
    • Identifying causes of high bills
    • Better customer service
  • Two-way communication with end-use loads for verification of load dispatch is not likely to happen, even though verification is really important for demand-side management programs.
  • It could be useful to learn more about the techniques used in Conditional Demand Analysis.
  • Load researchers do not think of sub-cycle statistics (e.g., harmonics) or other high-frequency information when talking about load research. They mostly think about hourly active power values.
  • Non-Intrusive Load Monitoring is alive and well, as I had thought.

Education and Learning, Today


Machine Learning class at

So last week I decided to enroll on Andrew Ng’s Machine Learning class from Stanford through the online education venture that he and others are working on: Coursera. I wanted to give this a try and better understand the benefits and disadvantages of this method. The class started today, and my first impression is: Wow! You mean anyone in the world can take advantage of this resource for free1 anywhere in the world? That’s amazing.

But you want to know what’s really amazing? The fact that I graduated from a doctoral program in a good engineering school in the US only two years ago and already I feel like the number and quality of the resources available to students today is immensely greater than what was available to me when I was a student. Yes, two years ago.

You know what else is amazing? That before I came to the US I was on a personal quest to educate myself through whatever means possible and at that time this meant: (a) obtaining digital copies of books that were unavailable at local libraries (through legal means or otherwise) and sometimes printing them, (b) downloading lecture notes and homework assignments posted on the web and found via search engines, (c) spending part of my salary to subscribe to at least one academic journal2, etc. If these kinds of resources were available to me at the time, I have little doubt that I would have been made use of it and who knows what effects it would have had on me and my life.

I still clearly remember what I felt during orientation day when I got to CMU in 2006 and the representative from library services described the resources that were available to all students. My jaw dropped. You mean, all that information is there for us to use?

I still don’t fully understand what effect that had in me, and it will definitely take me some time to grasp what it means for me and my students to have access to full courses from top schools. But I guess that is why I signed up for the class. Let’s find out.

1 Free may be misleading, given that you need at least the following: access to a computer, a high-speed Internet connection, knowledge of the English language as well as basic science and engineering.
2 The cost of this has still not changed, unfortunately.

Non-Intrusive Load Monitoring: Alive and Well in 2012

Start-up transientWe are only two and a half weeks away from the NILM Workshop that I am organizing along with Zico Kolter, and the response so far has been very positive. So much so, that I have had to readjust my preconceptions about how much interest there is for this topic out there.

We have almost 50 confirmed participants from a wide variety of places and backgrounds. Being this a first-time, half-day workshop on such a specialized topic, my expectations for attendance were not too high. I never expected that I would have to close registration to the workshop before the event took place, but yesterday we had to do this.

Similarly, I thought I knew about the majority of the people who were working on this problem and I was wrong about that, too. There is active interest in this topic from utilities, government laboratories, international agencies and private businesses. Just five days before our workshop, for instance, I am going to be speaking at an event organized by EPRI, with similar objectives as our workshop and targeted to utility personnel, state & federal regulators and staff, government researchers and policy makers.

The number of companies that are built almost entirely around the idea of extracting useful information out of electricity consumption measurements seems to be growing. At least five of the talks on our workshop will be from industry participants.

Is the time truly ripe for NILM research? Or am I just biased? Either way, I think that the workshop will likely be a fun event.

Big Buildings, Big Data – Panel at the BuildingsNY Show

BuildingsNY, May 3rd, 2012

BuildingsNY, May 3rd, 2012

On May 3rd, I will be participating as a panelist at the BuildingsNY show, discussing the role of real-time data analytics on energy management practices. I will be joined by Jonathan Spitz, from Johnson Controls, and the panel will be hosted by Mark DeSantis from kWantera.

For participants, the expected learning objectives are:

  • Understand what the ‘Big Data’ phenomenon means for large consumers (>1MW) such as the extraordinary energy and sustainability possibilities inherent in different types, sources and uses of existing electricity, steam and gas data.
  • Learn why smart meters are not (yet) very smart and what you can do about it.
  • Learn how, what, and where the data you already generate (and probably discard) can improve your energy efficiency, your power quality and your energy procurement.
  • What does this mean for me cost/benefit-wise.

Here is a link to a page with more information about the session:

2nd KDD Workshop on Data Mining Applications In Sustainability

SustKDD 2012

Co-located with KDD 2012

When: August 12, 2012 (Papers due on May 9, 2012)
Where: Beijing, China
How: More information…

Besides co-organizing the NILM Workshop which will take place in early May, I am also helping to organize another workshop on Data Mining Applications In Sustainability, which will be co-located with the Knowledge Discovery and Data Mining (KDD) conference that will take place in Beijing from August 12-16, 2012.

The goals of this KDD workshop are to:

  • to bring together researchers working on applications of KDD to sustainability in diverse areas, especially in infrastructures such as IT, Smart Grids, water, and transportation. This will facilitate cross-disciplinary collaboration and cross pollination of ideas across domains.
  • to familiarize the mainstream KDD community with diverse application areas within
  • to serve as a meeting ground and launchpad to galvanize and foster the development of this budding sub-community.

If you are interested in participating, we are accepting submissions of 8, 4 and 2 page manuscripts on a variety of topics including Transportation, Information Technology, Water Infrastructure, Building Energy Management, etc. These submissions are due on May 9, 2012.

Please visit the website to obtain more information:

1st International Workshop on Non-Intrusive Load Monitoring

When: May 7, 2012
Where: Carnegie Mellon University, Pittsburgh, PA
How: More information…

One of the things that has been keeping me busy these days is a workshop that I am helping to put together on the topic of electricity disaggregation or, more technically, Non-Intrusive Load Monitoring (NILM). These words may be familiar to some of you, given that I have previously written about this topic here a few times, but if you are new to the term, the basic idea is this: given a set of measurements of voltage and/or current taken at a limited number of locations of the power distribution system in a building, can you automatically obtain estimates of the electrical consumption of each individual appliance?

Because this topic played a large role in my doctoral work, I have been getting acquainted with different researchers around the world who are also working on this problem. And in the past five years or so, there has been a renewed interest in this topic, perhaps due to the increased attention paid to cleantech (although apparently that moniker is dead), as well as the proliferation of smart meters. However, as far as I know there has been no formal attempt to bring together everyone working in NILM and discuss the future of this research area. That is precisely what I wish to do with this workshop.

Some of the things we will be discussing are:

  • Review the main types of approaches that have been explored to date
  • Discuss possible paths forward knowing what has been tried and what has yet to be experimented
  • Possible solutions to the growing need for standardized datasets and performance metrics (here is a post explaining this in more detail)
  • Opportunities for collaboration among different research groups

The event will take place on May 7, 2012 here at Carnegie Mellon University, Pittsburgh, PA.

If you are interested in participating, registration is free. We will be accepting 2-page abstracts for poster presentations and short talks.

So far we have confirmed participation from researchers at a few government labs like EPRI and PNNL; a number of universities (MIT, University of Washington, CMU, University of Southampton); as well as a good number of industry participants: Bosch Research, Samsung Electronics, HP Labs, Belkin, PlotWatt, Emme, etc.

You can find more details by visiting our website:

© 2014 Mario Bergés

Theme by Anders NorenUp ↑