Sunday, 13 October 2019

Computational Intelligence (CI) Dream

I had the weirdest dream...

It was about the Twitter platform, except instead of tweets, people were tweeting these little AI programmes that represented the point you wanted to make.

The tweets would go out and then either interactively represent your argument in a discussion, or do other interactive things, like dance in time to music 'client side'.

For some reason people were calling them 'Computational Intelligences'. I can only assume this was out of respect for the derogatory connotation of 'Artificial'.

So you had this tweet stream of CIs that essentially talked to you, tried to engage you, or did other things in response to you.

The next phase were these AI documents. It was like, given that we can produce an infinite amount of text now, using language models, the actual text itself is irrelevant. Once there is essentially infinite text, any given text, no matter how persuasive or novel or important, is essentially worthless.

Instead we have 'smart documents', that embody the essence of the argument, I guess using finding from computational argumentation.

So you'd draft your argument, rather than a specific text, and the CI would go out and advocate your argument.

The other thing was it could produce the best text for a given argument, allowing you to re-structure your argument and automatically generate a text for it. Like a context sensitive drag and drop. Moving this concept into the introduction changes the font, or the form, of the argument.

Well... it was just a dream.

Monday, 9 April 2018

AMFIE correspondent bank accounts

For some reason, this information isn't available anywhere on the web... It is a list of the AMFIE correspondent banks (accounts you can use to transfer funds into your AMFIE account, and the associated currency of the account).

AMFIE Correspondent banks (PDF)

To send money to one of these accounts, just include your AMFIE account number in the reference information on the SEPA transfer.


Correspondent banks
Accounts you can use to transfer funds

Banques correspondantes
Banques où vous pouvez effectuer vos virements


Country
Pays

Bank
Banque

Currency
Devise

AMFIE account number
Numéro du compte de l’AMFIE

Swift/BIC code
Code Swift/BIC

Austria
Autriche

Bank Austria
Agence V.I.C.,
A-1400 Vienna

EUR AT33 1100 0093 7023 3000 BKAUATWW

Australia
Australie

Commonwealth Bank of Australia
48 Martin Place
Sydney NSW 2000

AUD 14632785

BSB number: 062000 CTBAAU2S

Belgium
Belgique

ING Belgium S.A.
Champ de Mars 23,
B-1050 Brussels

EUR BE69 3101 4150 2878 BBRUBEBB

Canada
Canada

Toronto Dominion Bank
Beaver Hall & Viger
525 Viger Ouest Suite 100
Montréal, P.Q., H2Z 0B2

USD
CAD

0010-7302531 - Transit 41601-004
0010-5205809 - Transit 41601-004 TDOMCATTMTL

France
France

ING Bank France
Immeuble Lumière
40 avenue des Terroirs de France
F-75616 Paris, Cedex 12

EUR FR76 3043 8000 0141 7790 3600 017 INGBFRPP

Germany
Allemagne

ING-DiBa AG
Hamburger Allee 1,
D-60486 Frankfurt

EUR DE36 5002 1000 0010 1388 08 INGBDEFF

Italy
Italie

ING BANK N.V., Milan Branch
Via Arrigo Boito, 10
20121 Milano, Italy

EUR IT55 Z034 7501 6010 0005 2092 410 INGBITMMICM

Luxembourg
Luxembourg

Banque Internationale à Luxembourg
(B.I.L)
69, route d’Esch,
L-2953 Luxembourg

EUR
USD
CHF
GBP
CAD
AUD

LU63 0023 1723 2326 7300
LU58 0023 1723 2329 9400
LU93 0023 1723 2325 9000
LU41 0023 1723 2327 0800
LU47 0023 1723 2327 6900
LU67 0023 1723 2326 4900

BILLLULL

ING Luxembourg S.A.
26, Place de la Gare
L-1616 Luxembourg

EUR
USD
CHF
GBP
CAD
AUD
DKK

LU35 0141 6408 6620 0000
LU30 0141 4408 6620 3010
LU93 0141 2408 6620 3050
LU17 0141 8408 6620 3030
LU08 0141 5408 6620 3110
LU48 0141 2408 6620 3860
LU63 0141 3408 6620 3100

CELLLULL

Netherlands
Pays-Bas

ING BANK N.V.
P.O. Box 23432, 1100 DX Amsterdam
Location code DEA 03

EUR NL55INGB0007735750 INGBNL2A

Spain
Espagne

ING BANK N.V., Sucursal en España
C/Genova, 27, 3rd floor
28004 Madrid

EUR ES78 1465 9000 1700 0186 3912 BBRUESMX

Switzerland
Suisse

ING Belgium, Brussels, Geneva Branch
6 rue Jean Petitot,
CH-1204 Geneva

CHF
EUR

CH84 0838 7000 0010 80473
CH79 0838 7000 0011 80473 BBRUCHGT

United
Kingdom
Royaume-Uni

Barclays Bank Plc.
31 Market Place,
Wokingham, RG40 1AR

GBP GB41 BARC 2011 7440 0556 46 BARCGB22

Thursday, 17 March 2016

Stop Scheduled Emails from Google Analytics

Possibly the most hidden feature ever? Not to mention changing every release... The information below refers to the version of GA as of March 2016 (also known as "the new-new-version").

Summary:
Admin > VIEW > Scheduled Emails

Details:
  1. Click "Admin" from the menu at the top.
    1. Select the right "ACCOUNT" from the Account column.
    2. Select the right "PROPERTY" from the Property column.
    3. Select the right "VIEW" from the View column.
  2. Select "Scheduled Emails" from  under the View column.
    1. Below the "PERSONAL TOOLS & ASSETS" section!
  3. Now scroll back up. (Yes, it's that well hidden!)
  4. Select "Actions".
  5. Select "Delete".
  6. Confirm.

Dear Google! Was this designed by a Microsoft engineer?

May I suggest that the email itself should contain a link that takes you directly to step 3 (assuming you have the right privileges to see that page) or a simple 'unsubscribe' link for users that don't.


kthxbi

Wednesday, 2 December 2015

Dawn of a new age or return to the Dark Ages?

Dear Drs. Xxxx and Xxxxx,

We write to alert you to a recent publication in Nature Genetics in which sequence data do not appear to have been correctly submitted to the public databases of the International Nucleotide Sequence Database Collaboration (http://www.insdc.org) in accordance with established best practice in the scientific community, or were missed to be cited as such.

We wish to remind you of the immense immediate and long-term value of having data deposited in the permanent public databases of record and of the significant convenience for the readers of Nature Publishing Group to be able to access systematically the data that underlie the studies about which they are reading. In addition, we refer you to the 'availability of data, material and methods' page and the 'Mandates for specific datasets' section of your instructions to authors in relation to availability of nucleotide sequences associated with publications:
http://www.nature.com/authors/policies/availability.html#data

In Xxx Xxxx et al., there is reference to the genomes of xxxxxxxxx Xxxxxx xxxxxxx varieties xxxx and xxx and a wild xxxxxxxxx relative Xxxxxx xxxxxxxxxx variety xxx. Associated RNA-Seq raw reads as well as genome assembly and annotation are all reported to be available from a non-INSDC website.

We encourage you as a lead author to provide any missing data from INSDC databases that is associated with this publication and recommend that Nature Publishing Group ensures data submission to one of the databases of the INSDC prior to publication and require data release no later than the time of publication, with full citation of INSDC-issued accession numbers in the manuscript.


Sincerely,

Xxxxx Xxxx on behalf of International Nucleotide Database Collaboration

Thursday, 26 March 2015

Official Gmail Blog: Hide morning and night hours in Calendar

Official Gmail Blog: Hide morning and night hours in Calendar



Nice except you can't see any information by hovering over a "compact" event! The only way is to click on the left to change the range, hover over the event to get the summary, then 'change the range' by sliding the hidden range back and forward one hour. Clicking or hovering over the "compact" event should bring up information just like normal.

Friday, 15 August 2014

Installing counterpartyd

There are some docs out-there to talk you through it (README.md), but it felt like a bit of a pain... Here is my 'concise' version with notes...

I'm running Mint on wheezy/sid
Kernel: 3.11.0-12-generic x86_64 (64 bit) Desktop: Gnome Distro: Linux Mint 16 Petra

BTC

I already had bitcoind up and running (it came with the distro), i.e. see:
apt-cache search bitcoin

(or http://counterpartyd-build.readthedocs.org/en/latest/SettingUpBitcoind.html)

In my case this installed the blockchain under ~/.bitcoin (I'm not sure if I configured that, or if that's where bitcoin-qt puts it by default).

In that location (~/.bitcoin), here is what I have in bitcoin.conf:
rpcuser=bitcoinrpc
rpcpassword=whatever99
txindex=1
server=1

Note that bitcoind is a daemon you run on your own machine (hence the term 'node'). The daemon accepts remote procedure calls (something like API calls), for example:
bitcoind getblockcount

As far as I understand, the rpc user and password settings are what the daemon uses to restrict such calls. As counterpartyd needs to talk to bitcoind, we'll need to pass that password to counterpartyd later.

The tindex line indexes the blockchain as it's downloaded, and is required for functionality used by counterpartyd.

The server line allows bitcoind to receive rpc calls.

You can add testnet=1 to run on the bitcoin testnet.

While running bitcoind, the network, memory and CPU usage of my computer went through the roof. This is expected, as the bitcoin blockchain is first downloaded, verified and indexed. It's a one off thing that settles down once you're back at 1 block every 10 minutes. Note that bitcoin-qt does exactly the same thing... literally. i.e. don't run bitcoin-qt at the same time as bitcoind, because they both try to download the blockchain to the same location!

XCP

I grabbed counterpartyd from git and switched to the (currently) latest tag:
git clone  https://github.com/CounterpartyXCP/counterpartyd.git
cd counterpertyd; git checkout  v9.39.0

I installed all the python requirements using virtualenv like this:
virtualenv MyVE --python python3
source MyVE/bin/activate
pip install --upgrade -r pip-requirements.txt

I found that I needed to use apt to get some development libraries:
sudo apt-get install python3-dev

Running ./counterpartyd.py --help showed that I also needed apsw. Trying to install that with pip failed due to first missing and then out of date libsqlite3-dev.

So... outside the virtualenv (for no good reason) I grabbed apsw from git:
git clone https://github.com/rogerbinns/apsw.git

Fetched the latest sqlite3 (using my virtualalenv):
~/counterpartyd/MyVE/bin/python setup.py fetch --sqlite --missing-checksum-ok

And finally installed apsw (using my virtualalenv):
~/counterpartyd/MyVE/bin/python setup.py install

Now at least ./counterpartyd.py --help worked, but figuring out the right settings for 'rpc' was a pain... The documents refer to '--bitcoind-rpc-password', which I dutifully set in a file called counterpartyd.conf. Naturally, I set it to match the rpcpassword in bitcoin.conf. However, running something like ./counterpartyd.py server (which is what you should run to start the counterparty daemon) complained about --backend-rpc-password not being set, which was confusing.

It seems that the 'backend' is bitcoind, so I should have done this:
./counterpartyd.py --backend-rpc-password=whatever99 server

Confusingly this then complains about --rpc-password not being set. You can just set this to 'whaever101', as it's just the counterpartyd equivalent of the bitcoind rpc-password. 

So to be clear (because it confuses me):
  • bitcoind uses something called rpcpassword to restrict access, and you set it to whatever you want when you start the deaemon (e.g. whatever99).
  • counterpartyd needs to know that password, but it calls it backend-rpc-password
  • counterpartyd uses something called rpc-password to restrict access, and you set it to whatever you want when you start the deaemon (e.g. whatever101).
As far as I could tell everything I tried setting in counterpartyd.conf was ignored. All the above was confusing to work out because of verbose error messages and confusing 'connection' errors when you got things wrong.

So finally:
./counterpartyd.py --backend-rpc-password=whatever99 --rpc-password=whatever101 server

The first thing this does (analogously to bitcoind) is parse through the blockchain and pull out all valid counterparty 'actions' into a database.

It's fun to watch the burn reply:
Burn: 16kNtJqcAqKDgpTGor2jYjahxLn1khiony burned 1.0 BTC for 1467.72727273 XCP (ef597b82df3afde4236f0d0e032160d02af31ac96cf28dc10217fb73a47987ef) [valid]

It's strange to think that setting fire to your money could be a sound investment...

Well... Now I'm stuck waiting for the index to build...