Profile

Cover photo
Gunnar Aastrand Grimnes
Worked at DFKI GmbH
Attended University of Aberdeen
337 followers|179,048 views
AboutPostsPhotosYouTubeReviews

Stream

 
There seems to be a bug when I combine non-ascii characters, quotes and the OR operator when searching. 

All of these queries work ok: 

"pølser"
"pålegg"
pølser OR pålegg
"egg" OR "spekemat"

but

"pølser" OR "pålegg" 

always returns no result. 

Any ideas? 

Cheers!
1
Matt Stein's profile photoGunnar Aastrand Grimnes's profile photo
2 comments
 
Thanks Matt! I didn't even know that site existed!
Add a comment...

Gunnar Aastrand Grimnes

Shared publicly  - 
 
 
AI startup Vicarious claims to have a system that can solve CAPTCHAs with "succes rate up to 90%".  

Beware: It's a textbook example of AI hype of the worst kind

Hype is dangerous to AI. Hype killed AI four times in the last five decades. AI Hype must be stopped.

Perhaps Vicarious can get "up to 90%" accuracy on some CAPTCHA dataset they cooked up, but
- (1) breaking CAPTCHAs is hardly an interesting task, unless you are a spammer;
- (2) it's easy to claim success on a dataset you cooked up yourself. There is no risk someone else will beat you. 
- (3) recognizing object in images is much, much harder than breaking CAPTCHAs. Some deep learning systems can already do this with decent accuracy. Some such systems have been deployed by Google and Baidu.
- (4) doing simultaneous segmentation and recognition of character strings is hardly a breakthrough. See demos of a 20 year-old system here: http://yann.lecun.com/exdb/lenet/index.html

The sad thing is that this announcement is being picked up by a number publications, including MIT Tech Review, Forbes, etc.

Here is an advice to scientific/tech journalists: please, please do not believe vague claims by AI startups unless they produce state of the art results on widely accepted benchmarks. 

This is particularly true for claims in image and speech recognition for which good benchmarks exists. For image recognition, a good example of such benchmark would be the ImageNet Large Scale Visual Recognition Challenge.

Whenever a startup claims "90% accuracy" on some random task, do not consider this newsworthy.  If the company also makes claims like "we are developing machine learning software based on the computational principles of the human brain" or uses impressive-sounding names like "Recursive Cortical Network", be even more suspicious. 

There are extremely impressive applications of deep learning out there (e.g. deployed by Google, Baidu, Microsoft, IBM, and a few startups), but this is not one of them.

Google's automatic photo tagger and Baidu's image retrieval system are much, much more impressive than the system in this announcement. Even if we just talk about challenging character recognition tasks, Google's system for picking out house numbers in StreetView images is way more impressive than this. 

AI "died" about four times in five decades because of hype: people made wild claims (often to impress potential investors or funding agencies) and could not deliver. Backlash ensued. It happened twice with neural nets already: once in the late 60's and again in the mid-90's.

Don't let it happen again. Beware of hype.

And by the way, no one is interested in breaking CAPTCHAs except spammers and computer security researchers. That's why you won't find many computer vision papers on the topic. That's also why it would be easy to break records, even if a standard dataset existed. 
1
Add a comment...
 
Is it possible to change the attributes of a type that is shown in the element-table on the entries screen in CP? I.e. I've got a "contact persons" channel and I'd like to show the "department" attribute in the table. 
1
Brandon Kelly's profile photo
 
Not yet, but it's on our list!
Add a comment...

Gunnar Aastrand Grimnes

Shared publicly  - 
 
Christmas can come to +Bakken & Bæck !
2
Add a comment...

Gunnar Aastrand Grimnes

Shared publicly  - 
 
We've just released python  #RDFLib 4.0! Best release yet!  #RDFa ,  #SPARQL1 .1 any pretty much any other feature you want! Also, brand new up-to-date documentation!
We've just released RDFLib 4.0 - and it's the best RDFLib release so far! We now have: * SPARLQL 1.1 support * RDFa/MicroData support (actually already in 3.4) * SPARQL(Update)Store integrated in core * Massively overhauled documentation: https://rdflib.readthedocs.org/ And many other changes!
3
2
Gunnar Aastrand Grimnes's profile photoIvan Herman's profile photo
Add a comment...

Gunnar Aastrand Grimnes

Shared publicly  - 
 
test
1
Gunnar Aastrand Grimnes's profile photoCédric Mesnage's profile photoWalter Kammergruber's profile photo
4 comments
Add a comment...
Have him in circles
337 people
‫השכרת רכב‬‎'s profile photo
Ahsan Morshed's profile photo
Suzi Kovač's profile photo
Simon Thompson's profile photo
Tim Zijlstra's profile photo
Quentin Reul's profile photo
john onyemaechi ugwoke's profile photo
Stefano Bertolo's profile photo
Chris Jackson's profile photo
 
Hi there, I seem to have some problems with updating asset indexes, I have a folder with about 2500 images and it was all indexed. Then I realised that some of the filename had various accents and other unicode characters in them, and I've renamed the files. BUT updating asset indexes doesn't find all the new files, nor does it detect all the files that are no longer there. 

Is there a trick to make the "update asset indexes" work better? Never change away from the browser tab where it "runs"? It takes quite a while for the 2500 files and it gets boring to watch it after a while :) 

(The progress indicator is also way off, it goes to 100% complete almost immediately and stays there for the next 45 minutes, although inspecting the http calls being made it has the right offset/total count)

Cheers!
1
Gunnar Aastrand Grimnes's profile photoBrad Bell's profile photo
3 comments
 
Thanks, Gunnar... will look into it!
Add a comment...
 
How do I set relationships programmatically from a plugin? 

I've got a plugin that imports some data intro entries, the entry has an "image" field, which should be set to one or more assets. The assets already exist, and I can find the one I want with craft()>assets>findFile(array( 'search' => 'blah' )) - but I cannot seem to set the "image" field.  

When saving my entry I get a big ugly: CDbCommand could not execute SQL command: SQLSTATE[01000]: Warning: 1265 Data truncated for column 'targetId' at row 1. The SQL statement executed was: INSERT INTO `craft_relations` (`fieldId`, `sourceId`, `targetId`, `sortOrder`, `dateCreated`, `dateUpdated`, `uid`) VALUES (:row0_col0, :row0_col1, :row0_col2, :row0_col3, :row0_col4, :row0_col5, :row0_col6). Bound with :row0_col0='64', :row0_col1='16928', :row0_col2=Craft\AssetFileModel::__set_state(array(
'elementType' => 'Asset',
'_transform' => NULL, ... error 

Any ideas? 
1
Brandon Kelly's profile photoGunnar Aastrand Grimnes's profile photo
4 comments
 
ah - looking at the asset-uploader-plugin that was posted here gave me the answer: https://gist.github.com/boboldehampsink/7211614#file-assetuploadservice-php-L73

You must put just the asset->id in the array to set, not the whole object. 
Add a comment...

Gunnar Aastrand Grimnes

Tips, Tricks, and Tutorials  - 
 
I just discovered some funny behaviour with multi-select options in twig. It seems the multi-set options do not compare equal to twig strings, at least for the containment operator: 

{% for value in myField.options %}
{% if value in ['a', 'b', 'c'] %}
we never get here
{% endif %}
{% endfor %}

Adding {% set value = value ~ '' %} above the if fixes it - but it's hard to know and a bit ugly ;) 
1
Brandon Kelly's profile photoGunnar Aastrand Grimnes's profile photo
2 comments
 
I saw that when I dumped it - but somehow twig wouldn't let me do "if value.value in [ ... ] " - anyway - it took me some time to work out - so I left it here for others to find!
Add a comment...

Gunnar Aastrand Grimnes

Shared publicly  - 
1
Add a comment...

Gunnar Aastrand Grimnes
owner

Discussion  - 
 
 
We've just released python  #RDFLib 4.0! Best release yet!  #RDFa ,  #SPARQL1 .1 any pretty much any other feature you want! Also, brand new up-to-date documentation!
We've just released RDFLib 4.0 - and it's the best RDFLib release so far! We now have: * SPARLQL 1.1 support * RDFa/MicroData support (actually already in 3.4) * SPARQL(Update)Store integrated in core * Massively overhauled documentation: https://rdflib.readthedocs.org/ And many other changes!
4
Add a comment...

Gunnar Aastrand Grimnes

Shared publicly  - 
 
my G+ post that just said "test" is my most commented on post so far!
1
Florian Mittag's profile photoSven Schwarz's profile photo
3 comments
 
This is a comment from my phone! Yay!
Add a comment...
People
Have him in circles
337 people
‫השכרת רכב‬‎'s profile photo
Ahsan Morshed's profile photo
Suzi Kovač's profile photo
Simon Thompson's profile photo
Tim Zijlstra's profile photo
Quentin Reul's profile photo
john onyemaechi ugwoke's profile photo
Stefano Bertolo's profile photo
Chris Jackson's profile photo
Work
Occupation
Researcher, DFKI, Kaiserslautern
Employment
  • DFKI GmbH
    Senior Researcher
Basic Information
Gender
Male
Story
Introduction
I am a semantic web/machine learning researcher in Kaiserslautern, Germany.
Education
  • University of Aberdeen
    Computer Science, 1999 - 2007
Quick, friendly, fair price!
Public - 2 years ago
reviewed 2 years ago
1 review
Map
Map
Map