Entries : Category [ Computers and Technology ]
[Miscellaneous]  [Computers and Technology]  [Travel]  [Education]  [Hacks]  [Robotics]  [Science]  [Programming and Software]  [iPhone]  [Digital TV and Video]  [Intellectual Property & Copyright]  [Personal] 

18 September

Today I gave a one-hour talk at the IEEE 9th International Symposium on Robotics and Sensor Environments (ROSE). I talked about our work in underwater robotics (as I often do) and specifically dealt with summarization, terrain identification and a tiny bit about human-robot interaction.

One of the components of the talk was work by my former student Philippe Giguere, now a professor and Laval University in Quebec City. He was in the audience which made that part of the presentation feel a bit weird to me -- here I was reminiscing about work he had done with me and almost sharing the odd inside joke with him while this audience was listening. It was a very nice audience and they asked a lot of good questions, which left me with a generally positive vibe. Not a bad way to start a Sunday morning, after all!

The key coauthors of the work I presented, in addition to Philippe, are Yogesh Girdhar and Junaed Sattar.

The summarization part of the talk is covered (at a very high level) in my TEDxMcGIll talk shown here.

By Gregory Dudek at | Leave a comment |    
21 September

Quote of the day (Montreal graphs):

To compare the performance of exploration and validation, both algorithms were tested on a variety of random graphs. The first set of parameterized random graphs was generated by starting with a complete 2D lattice (i.e. a grid) and deleting a specified fraction of randomly selected edges such that the graph remains connected. This first family of graphs should be familiar to those who have been forced to drive a car in Montreal (where roads are often under repair in the summer), and are termed Montreal graphs with deletion factor p, or Montreal(p), p ∈ (0,1).

-- from a paper in 1997. I think maybe the term was first used in 1994.

These can also be described as geometric graphs generated by the Erdős-Rényi process,
constrained and to the 4-connected lattice and with a side condition to
maintain connectivity.

I put this first on my Google Plus stream, so I guess this counts as a cross-post.

A montreal graph

with a large number of deletions (large p).

By Gregory Dudek at | Read (1) or Leave a comment |    
08 January

Computer science is becoming increasing popular again, after a huge upswing in the 1990's, a lull after the dot-com meltdown in 2001 and a gradual growth since then. In the last few years computer science departments all across the US and Canada have been seeing enrollments grow, and i the last year or two they have ben growing quite a lot. At McGill, we have seen a steady but gradual growth in enrollment for a few year, but it is now becoming a surge.

In the last few days, there are been some notable popular manifestations of this trend. New York City mayor Michael Bloomberg made a New Year's resolution to lead to program and posted it on Twitter:

My New Year's resolution is to learn to code with
Codecademy in 2012! Join me. codeyear.com #codeyear

A major salary survey is about to be released that shows CS salaries have been rising (when many other sectors are not seeing growth) and "Midsize and large companies are both aiming to hire more IT pros. The midsize are seeking IT executives (such as VPs of information services and technical services), as well as programmers, database specialists, systems analysts, and voice/wireless communication pros. Enterprises are moving IT and data center operations back in-house, which means greater demand for data center managers and supervisors."

By Gregory Dudek at | Leave a comment |    
05 February

The effect and influence Facebook is pernicious and profound. What makes this web property especially damaging is that the harm that it does to the fabric of our culture is subtly concealed. The promise of the Internet, and its most obvious incarnation the World Wide Web, is that it democratizes information dissemination and gathering. Every person becomes a potential source of creative output. Whether it is political journalism, how to improve one's garden, the nature of the universe, or the subtleties of traveling through Europe, every person can collect their thoughts and can post their opinions and thoughts and reflections. The most obvious effect of this is to make far more raw information available to people on a timely basis. That's great, but it's only one side of the coin, and the other represents an equally critical step for our society. By allowing anybody to post their thoughts in the hope of helping others, it engages far more people than ever before in the creative process. This in turn, gets these people to think more deeply about their world, and about how they can contribute to it.
What's key here it that the web pushes people to think more deeply and be more creative, but that outlet and effect is very dependent on people believing that many people may find their information and that they can use plenty of words and space to describe their thoughts.

The problem with Facebook, is that by creating a password-protected space in which users interact, it prevents the critical access to information that makes the Internet possible. Information that users post in Facebook simply isn't out there for the whole Internet to see. More pernicious, and less obvious, it focuses people's attention increasingly on their day-to-day interactions with their friends, or the banal hijinx of their pseudo-friends, and focuses the creation of content, on petty issues that would only appeal to one's small circle of intimates. On top of that, information is always compressed into little bite-sized chunks. The nature of the discourse is more like a chat that a friend suffering from attention deficit disorder than an article, encyclopedia entry, or even one's personal diary. Other social networks such as Google plus, also suffer from this to some extent, but because Facebook is unusual in both its penetration and the manner in which it seals off its internal content from the rest of the Internet, it represents a particular low point. While twitter forces information into little dribbles, it remains much for open and is not likely to displace the creations of web sites or blogs, but drive traffic to them.

On top of that, there is also the issue of Facebook owning (and having access to) you email, photos and other personal information. This is not a concern unique to Facebook, but it's just one more reason not to love it, and it is just an invitation to abuse. Furthermore, Facebook has not been the most impressive net citizen in terms of open access, control over one's personal data, or the consistency and fairness of its privacy policy.

Now with the recent Facebook IPO, not only will Facebook be propelled into greater visibility, but more people will be trip be contracted to this kind of sad, dismal, descriptive business model.

One positive note for those who dislike Facebook, is that the attempt to aggregate large amounts of information into a single web property is universally led to failure. Yahoo was perhaps the most successful doing this and even day did not succeed particularly well. It is not clear that the combination of substantial information properties, such as newspapers, with the comings and goings of one's personal friends can be merged successfully into a single concise web presence. Furthermore, based on Facebook's initial attempts to do this kind of thing, there is little evidence that they're the ones who can pull it off. On the other hand the power and influence is immense and with large amounts of capital, they will have huge incentive for you to push beyond the limits of what they can do well today. Here's to hoping they fail dismally.

[ This article was generated using a new dictation technology. I am still stamping out all the strange misinterpretations. ]

[ More recent: A recent talk/article by Eben Moglen of Columbia law school also brings up the concerns with Facebook messaging. ]

By Gregory Dudek at | Read (2) or Leave a comment |    
10 April

Google announced their project Glass in early April. It's a head's up display concept project for letting you link the internet and the kind of experience you get on a smart phone with your normal visual experience.

The general concept of a head's-up assistive display has been around for a quite a while, both in the scientific community as well as in the science fiction world. Pattie Maes from the media lab presented an impressive TED video of this kind of idea as implemented by Pranav Mistry. This was a demo of a real device, although it was also not ready from a consumer audience.

It any group or company can bring an idea like this to market, and make it work, it's Google. There is no way, however, that it will be flawless and like the Apple Newton, there is a risk that even if it is novel, cool, delivers something unique and pushes the edge, it will fail to gain wide approval due to the inherent glitches it will doubtless contain. A technology like this has many little problems to overcome, like the fidelity of
speech input and the calibration and positioning of the display. I want one, but I don't expect flawless perfection. Many consumers will be put off by a device that only "kind of" works, and thus Google might wisely, but unfortunately, defer release of this product for a long time. I hope not, but I wouldn't blame them.

By Gregory Dudek at | Leave a comment |    
12 May

In summer, my web server is at risk of overheating. Today I installed some temperature reporting and recording tools.

For my ubuntu linux, I am using the lm_sensors kernel module. This was installed auptmatically (Ubuntu 10.4).

To report temperatures, the "sensors" command can be used, but first the correct kernel drivers must be loaded. This can be done really easily, and with minimla prior knowledge, by using the command "sensors-detect". All the following
probably needs to be done as super-user.

# sensors-detect

sensors-detect revision 5818 (2010-01-18 17:22:07 +0100)
System: To be Filled To be Filled
Board: Supermicro Inc. Intel 440BX/440GX

Some south bridges, CPUs or memory controllers contain embedded sensors.
Do you want to scan for them? This is totally safe. (YES/no):
Silicon Integrated Systems SIS5595... No
VIA VT82C686 Integrated Sensors... No

After following a lot of simple instructions, I got the correct kernel module
loaded (in my case, a Winbond W83781D). Then I needed to start it by typing:

# service module-init-tools start

response: module-init-tools stop/waiting

# sensors

Adapter: ISA adapter
in0: +2.05 V (min = +1.79 V, max = +2.19 V)
in1: +2.06 V (min = +1.79 V, max = +2.19 V)
in2: +3.39 V (min = +2.82 V, max = +3.79 V)
in3: +2.94 V (min = +0.13 V, max = +0.00 V) ALARM
in4: +3.20 V (min = +0.00 V, max = +0.00 V) ALARM
in5: +3.34 V (min = +0.00 V, max = +0.26 V) ALARM
in6: +3.30 V (min = +0.00 V, max = +0.00 V) ALARM
fan1: 3125 RPM (min = 21093 RPM, div = 8) ALARM
fan2: 4115 RPM (min = 21093 RPM, div = 8) ALARM
fan3: 0 RPM (min = -1 RPM, div = 2) ALARM
temp1: +60.0C (high = +0.0C, hyst = +0.0C) ALARM
temp2: +63.0C (high = +70.0C, hyst = +65.0C)
temp3: -48.0C (high = +80.0C, hyst = +75.0C)
cpu0_vid: +2.000 V

Finally, to record this data and make it accessible via a web interface, I installed munin with the sensors plugin.

Munin works by running on a main (web) server machine, and then contacting munin node processes on each machine being monitored. In my case, the node is configured to include the sensors plugin, which is set up by creating a symbolic link (via ln) from the actual plugin at /usr/share/munin/plugins/sensors_ to a link called sensors_temp (to have munin monitor the temperature). The link command, in my case, was:

ln -s /usr/share/munin/plugins/sensors_ /etc/munin/plugins/sensors_temp

Testing this can be accomplished with munin_run:

# munin-run sensors_temp
temp1.value 60.0
temp2.value 63.5
temp3.value -48.0

For background info, see: http://www.lm-sensors.org/wiki/FAQ/Chapter2

By Gregory Dudek at | Leave a comment |    
Prev  1   2   3   4   5   6   7   8   9   10   [11]   12   13   14   Next