Archive for the ‘Technology’ Category

Australian food emojis

Tuesday, 4 June, 2019

A quick survey shows no fewer than 10 different emoji for specific items of Japanese cuisine (as of 2019’s Emoji 12.0 list). Which is fine, but clearly other cuisines are under-represented.

So I propose they add emojis for:

  • Chicken parma
  • Chiko roll
  • Fairy bread
  • Hamburger with beetroot
  • Lamington
  • Meat pie*
  • Musk sticks
  • Pavlova
  • Vanilla slice
  • Vegemite

* There is a pie emoji, but most people seem to interpret it as a sweet pie, and the glyph is actually a slice of sweet dessert pie on some platforms, including Google Android and Twitter. Which is clearly unacceptable and un-Australian.

Doing commercial research

Tuesday, 9 May, 2017

I do scientific research in my job, but I don’t work in academia. I work for a company, and my research goes into commercial products. If I was in academia, I’d be publishing my work and presenting it at conferences all the time. Working for a company, things are rather different.

I have just finished (yesterday) writing a presentation that I hope to give at the Electronic imaging 2018 conference in San Francisco, in February 2018. If accepted by the conference, this will be my first conference presentation since 2014. The deadline for submission of my presentation is 15 August, so I’ve completed it more than 3 months in advance. And I haven’t just completed the slides; I’ve also written a talk script, which will be almost word for word what I say during the presentation. And in the time between now and February next year, I won’t be able to change either the slides or the talk script.

The reason for this is that my talk has to go through the company intellectual property and legal department, to make sure that I am not disclosing anything which is a corporate secret. Besides writing my slides and script, I also had to highlight everything in my talk that was about scientific or technological knowhow from my research, and cross-reference it to published patents that I have written. This is to make sure that I don’t disclose anything that the company hasn’t protected in a patent application. Note that this is published patents. The patent office publishes patent applications 18 months after the filing date. So there could be some technology for which we filed a patent a year ago… and I wouldn’t be allowed to talk about it.

The process of vetting and approval by the IP and legal department takes up to 10 weeks. Thus the need for me to finalise my talk nearly 3 months before the conference submission deadline. Assuming my talk passes the IP/legal checks to make sure I’m not disclosing anything we haven’t got protected by a published patent, the company then still has to decide if it wants to let me talk about my work, or if it would simply rather keep it a secret. (At least, as secret as it can be if it appears in a published patent. It makes no sense to me, but yes, sometimes they quite specifically do not want you to talk about work that is disclosed in a published patent application.) This is the gamble phase – it’s impossible to know if the company will approve or reject the application to publish at this stage. So the only way to find out is to go to all the effort of writing your publication and putting it through this process. I could very well have wasted the past two weeks at work writing my complete, finished talk, hoping to present it, and be shot down at this stage. The same would go for a journal paper.

Speaking of journal papers, I am going through this long process for the first time in four years because Electronic Imaging has started accepting “oral presentation only” papers. These are delivered as talks at the conference, but do not appear as printed papers in the conference proceedings. They did this specifically to allow industry researchers such as myself to give talks. Because if I had to submit a written version of my work to go into the conference proceedings (as I did 4 years ago), it has to go through the same external disclosure approval process as the talk slides and script. This makes it much more difficult to publish, because as a printed paper it goes through anonymous peer review. And peer reviewers often request small additions or clarifications to the paper before they agree it is suitable for publication. That can be the kiss of death for my paper, because (a) I often don’t have additional time to devote to revising a paper, particularly if they ask for additional experimental results, and (b) the referee’s request may involve needing to disclose further potentially secret information. I have managed to pull all of this off before, but after the difficulty I had in 2014 getting my previous paper through, I had all but given up, until Electronic Imaging instituted the “oral presentation only” papers.

Anyway, assuming all of the IP/legal and corporate secrecy checks are passed and the company is okay with me disclosing the aspects of my research in my talk script, I can submit it to the conference! And if it’s approved, I can give the talk next February.

As I mentioned, I’m not allowed to change the content of my talk between now and then. If I do more research which improves on the results I plan to talk about, or which solves one or more of the outstanding problems before February, I won’t be allowed to mention it. If I’ve already done work now which improves on the results in my talk, but which I couldn’t include because it’s covered by a patent that has been filed but will not be published by the conference submission deadline, I’m not allowed to talk about that either.

This was in fact the situation for my last talk in 2014. I presented some work, and at the end I mentioned the elephant in the room: the most obvious problem with the results. I had already solved that problem and filed a patent describing the technique. It’s a beautiful piece of science and I am incredibly proud of it – it is I think the best piece of science I’ve done at this job. But the patent was not published by the time that talk had to be submitted to the conference (in mid 2013), so I had to leave it out. And then during the talk I had to raise this very obvious problem as an “unsolved problem”, and stay mum about the fact that I had already solved it over a year earlier. And hope that nobody in the audience pointed out the same solution during question time after my talk! (They didn’t – thankfully it’s not an obvious solution. It took me a lot of effort to come up with it, prove it worked, and solve the secondary problems it raised.)

Anyway, if my current application for the talk next year, in 2018, is approved by my company, I will finally be able to talk about the solution to that problem to an audience of fellow researchers. A solution I came up with in 2012.

I’m not sure I could survive in academia with its “publish or perish” mindset. I don’t have the workaholic temperament necessary to do well there. I’m happier in my corporate research job. But this approach to publication and dissemination of my work is incredibly frustrating. I have good work that I’m proud of and want to share with fellow experts in my field, and I want to establish a reputation as a quality researcher, but I have to jump through these multifarious hoops to do it.

So, if you’ve read this, I hope it provides some insight into the life and publication trials of a corporate research scientist.


Tuesday, 8 July, 2014

A friend suggested typing “List of A” into Wikipedia’s search box and seeing what it suggested, as a follow up to my previous auto-complete post. So without further ado, here are the top auto-complete suggestions for each letter:

One letter searches

Wednesday, 21 May, 2014

Just noodling around, I started typing something into Wikipedia’s search box, and before I thought of a second letter to type, it popped up a list of suggested auto-completions. I was slightly amused at the top hit, so decided to try every letter in the alphabet to see what I’d get. Here then is the list:

  • Animal
  • Bakhsh
  • Canada
  • Departments of France
  • England
  • France
  • Germany
  • Hispanic (U.S. Census)
  • Iran standard time
  • Japan
  • Keyboard instrument
  • List of sovereign states
  • Mollusca
  • New York City
  • Ontario
  • Poland
  • Quebec
  • Race and ethnicity in the United States Census
  • Spain
  • The New York Times
  • United States
  • Village
  • World War II
  • X
  • YouTube
  • ZIP code

This is a very interesting list. Presumably it is decided by some sort of algorithm running on Wikipedia’s servers that analyses the most popular search terms. Some of the entries are perfectly plausible on this assumption: England, France, Germany, United States, World War II. I can easily imagine they might well be the most popular Wikipedia search terms starting with those letters.

Others are more odd. Mollusca stands out a bit. Okay, I can see a lot of people might search for information on Animals in general, but molluscs? And by their technical name rather than common name? Iran standard time? Who on earth is searching specifically for information about the time zone of Iran? The two US Census entries are also intriguing… is there some reason for lots of people to look for information about race or ethnicity with regards to US Census taking?

The strangest one to me is Bakhsh. Until I tried this experiment, I’d never even heard the term. It’s an extremely specialised subject and the Wikipedia entry is quite short. Why is this showing up as the number one suggested search auto-complete beginning with B? I have no clue.

Trying the experiment further afield, here are the top auto-complete suggestions for typing one letter into Google’s search box (with explanations, since many of them are Australian companies – Google is clearly using my IP address to localise my results):

  • ANZ (ANZ Bank)
  • BoM (Australian Bureau of Meteorology)
  • Centrelink (Australian Government Department of Human Services)
  • dictionary
  • eBay
  • Facebook
  • Google
  • Hotmail
  • Instagram
  • JetStar (Australian airline)
  • Kmart
  • LinkedIn
  • maps
  • Netbank (e-banking service of the Commonwealth Bank of Australia)
  • Optus
  • PayPal
  • Qantas
  • real estate
  • SMH (The Sydney Morning Herald newspaper)
  • TV guide
  • UTS (University of Technology, Sydney)
  • Virgin
  • Westpac (another bank)
  • XBox
  • YouTube
  • Zara

What I’m working on

Friday, 18 January, 2013

I work for Canon Information Systems Research Australia, a subsidiary R&D company of Canon. It’s usually very difficult for me to say much about what I’m working on beyond “stuff to do with optics and cameras”, because of the need to keep our current research within the corporation. Once the work is patented and published, it becomes more public and I can point and say, “I did that” – but that’s two years or more after I actually did the work.

But I just found some stuff on Canon’s own public website that will give you an idea of what I’m working on right now. This. I’m working on this. Not all of it, just some aspects. But yeah, my work right now feeds directly into this, and more specifically the last section, about health management and safety.

Mac video

Friday, 28 December, 2012

So here’s my current summary of trying to find a video player for MacOS that will allow me to step frame-by-frame forwards and backwards:

  • VLC: Only steps forwards, not backwards. There is a Lua script which enables backward stepping, but after installing it, it doesn’t seem to do anything. I’m not sure if I’ve installed it correctly.
  • MPlayer MAC OSX: Only steps forwards, not backwards.
  • MPEG Streamclip: Apparently needs the Apple MPEG-2 Playback Component to be installed to get it to work. This is a $30 purchase from the Apple download store.
  • QuickTime: Step function both directions, but it is about 4 frames at once, not frame-by-frame.
  • Avidemux: Managed to get this working by following some online advice and deleting two of the library files in the install (since MacOS Mountain Lion apparently has those libraries already installed in the system). After some fiddling I finally figured out that on opening a video file it needed to write an index file, and for some reason I didn’t have permission to write into my video files directory. I gave myself write permission, and it loaded the file. But to display it at the right aspect ratio I actually have to set up a video filter and show the processed video. But once all that’s done, I can step frame-by-frame in both directions! Yay!

Mac enabled

Thursday, 27 December, 2012

I am up and running on my brand new iMac, after migrating from an old Windows XP machine. It’s taken a few days to get everything organised, files transferred, and various settings and things set up. I’ve transferred my Photoshop actions and styles, for example. iTunes took a bit of fiddling, but now seems to be okay. I’ve downloaded and installed a bunch of utilities like TextWrangler (a featureful text editor), FileZilla (FTP client), Inkscape (SVG editor), VLC (video player), and few other knick knacks.

I’ve now just set up a Time Capsule for backups and it’s making the first backup of the system and is working as my new WiFi router. Everything seems to be working nicely, without too much hassle.

The only thing I still need to confirm I can do with the new machine is make a new Darths & Droids strip from scratch. I used VirtualDubMod on Windows to play the ripped movie files, and it has single frame step forwards and backwards, which makes it easy to find the frames I want to screengrab for the comic panels. Unfortunately, VLC only does frame-by-frame stepping forwards, not backwards. (Google finds many people complaining about this fact and requesting step backwards as a feature, invariably followed by people saying the VLC developers refuse point-blank to implement it for some vague reasons.) I found someone recommending Avidemux as an alternative, which has frame-by-frame in both directions. So I installed that, but it crashes on starting.

Then I found someone had written a LUA script for VLC, which seems to do what I want from the comments. But I have no idea what a LUA script is or how to integrate it into VLC.

So I’m kind of stuck now. I really, really need a video player that will let me step frame-by-frame both forwards and backwards. Throwing this out there in case anyone has a solution.

EDIT: I found a guide to installing LUA scripts for VLC, and followed that. But it doesn’t seem to be working. I don’t see an extra step back function anywhere.

Hi-fi resolution

Saturday, 20 October, 2012

So, after some advice from here and elsewhere on my previously described hi-fi problem, I figured I’d try disconnecting the speakers one by one to see if I could isolate the problem. I loaded the offending Dr Who DVD, stuck the intro music on a repeat loop, and began fiddling.

First I reconfirmed the problem. The first thing I figured out was that it only occurred with the volume turned up above a certain level. Below about -30dB on my amp’s volume scale, no problem. Above that limit, the amp kept switching itself off at loud parts of the music.

Next I disconnected the left main speaker wires at the rear of the amp. Turned the volume up… and the amp stayed on. I reconnected the left speaker wires, turned the amp up… real high, to about -15 dB (way louder than we ever play anything), and it stayed on! I played with it a bit more to confirm the behaviour.

So, it looks like the wires connecting the left main speaker to the amp were somehow slightly unstable in their connection. Taking them out and reconnecting them seems to have completely solved the problem. I presume the wires were at some point close enough to be sparking or otherwise shorting when a loud bit of sound was being sent to the speaker, resulting in the amp circuit-breaking itself in self-preservation.

Yay! Thanks to all who suggested trying this.

Hi-fi problem

Tuesday, 16 October, 2012

Okay, I have a weird failure symptom of my audio gear. We upgraded our TV and speakers a few months ago, but still have the same DVD player and amplifier. A few nights ago we popped in a Dr Who DVD, planning to watch the Eccleston season again from the beginning. At a point in the first episode where there should have been an explosion (judging by the video), the amplifier spontaneously switched itself off. I switched the amp back on, rewound the DVD, and tried playing through the explosion again, 3 or 4 times, and the amp switched itself off at the same point each time.

Each time when I turned the amp back on, the front panel displayed an error message “CHK SPKR WIRES”, which I’ve never seen before, for a couple of seconds, before returning to normal working order. There’s no problem with the speaker connections as far as I’m aware – all speakers seem to be working fine. We watched the remainder of the episode, which also contained several more explosions, but the amp didn’t glitch at those.

We then watched a second episode on the same DVD, and it worked fine, until one point in the episode where the amp switched itself off again. There was no visual explosion on screen this time, and rewinding and playing through again caused the amp to switch off again at the same point, so I don’t know if there was a loud sound in the soundtrack at that point or not. Once past that point, the amp didn’t glitch again.

Last night we watched a DVD movie, without incident.

Just now, I put on the third Dr Who episode from the same DVD, and the amp is now turning itself off multiple times during the opening music/credits sequence. So annoying that we gave up and put the DVD away.

So now I’m wondering if it’s something to do with the audio levels on this Dr Who DVD, which might be overloading my new speakers and causing the amp to switch off in self-preservation, or if the amp is actually breaking down. We’ve watched this DVD before, with the same amp, but with older speakers, with no incident. I may have to do more experiments, but does anybody out there have any insights?

Australia to change top level domain name

Thursday, 2 August, 2012

BREAKING NEWS: Australia to change its top level domain name from .au to .ag, after winning substantially more silver than gold medals at London Olympics.