I’ve updated the JSNeuralNetwork library implementing a small and still simple version of the Hopfield Network, along with some implementations.
Moreover, I implemented a new experiment based on the Kohonen networks, which is about vision, and needs a webcam to work.
As usual everything is here: http://www.nicassio.it/daniele/JSNeuralNetwork/.
Today I ultimated the JSNeuralNetowork part which builds Feed Forward Neural Network. Now it only manages 2 layers network, with one input layer and 1 output neuron. It should be easy to extend them to a more general three-layer model, but my that was not my intent for now.
There’s an example here.
This is my last experiment: text which creates itself with a genetic algorithm, implemented in JS by my library JSGenetic, which I wrote some time ago and described in a previous post.
I used this kind of self-generating text to implement a timeline based curriculum vitae, of which I’ll have to review the content, but i’m quite happy with the structure.
As you can see when it computes the icons, this doesn’t work only for text, but with with generic pixel templates too.
These days I’ve been experimenting with Processing, and these are my very first results, ported to the web with Processing.js. I’ve done something slighly more complex, using some 3D and input from the webcam, which I can’t port to the web because the webcam/microphone input are not supported yet by Processing.js. Maybe I’ll post those sources later.
Today I made this chrome extension.
This semester at the university I attended a course of web programming, and as part of the final exam I was asked to develop a web application. It would use a Java servlet hosted by the Google App Engine, and would store data to the Google’s Datastore.
I thought that it would be much more useful to make a mobile web app, which is something slightly more challenging which I didn’t practice a lot in the past.
The app goal is to track daily expenses, by providing a way to insert, edit and delete transactions.
Of course the whole project was basically boring (which is different than easy), so I decided to add a funny functionality: I included a QR code scanner to add transactions by automatically, simply scanning the related QR code.
In the end, it has been fun, and as usual I’m writing this blog post to keep track of my works.
A live version of the app is hosted by Google here.
A simple test Microaccounting Store which provides QR codes to test the app is here.
Looking in the privacy settings of my Facebook profile, I discovered the option which indicates who can search you by the phone number you provided. The default setting is everyone.
Probably this option is intended for allowing users to match friends from the phonebook of their phones, but works in the desktop site too, by simply inserting the phone number in the searchbox. It works even if you have set your phone number to be private, it’s a different kind of option.
This is pretty similar to the kind of matching is made in WhatsApp or similar apps to find friends, but is completely different if you think that in those other services you cannot discover additional information like first and last name by providing the phone number, you just use the name you’re providing in your phonebook.
Using this Facebook feature lets you get a lot of information (Facebook’s public profile infos) by only providing the number of the person you’re interested in. That’s interesting.
Of course, to be searched this way, you need to have provided your number in some way to facebook. But this is more and more common, Facebook asks for it in various occasions, such as for the two step verification, and likely by using the mobile messenger app.
With that said, this service can be used to match unknown numbers, which is pretty useful [and a little creepy, if you think you can finally find out whose person number is being written in the toilette].
Interested by this possibility, I tried to brute-force the function to create an phonebook of random people from Facebook. I made a program which randomly tries a lot of numbers waiting for matches in the searchbox.
Actually, it worked pretty good, and I was able to retrieve like ~50 phone numbers and relative facebook profile by searching phone numbers similar to my own.
The program was set to run forever, but after a while Facebook noticed me that I was misusing their services, and asked me to enter a captcha to verify my humanity. 🙂
This method is probably not suitable for retrieving the phone numbers of every facebook user (actually I still think there’s a chance if you have a large number of fake Facebook accounts working simultaneously), but could be used by some spammers to find few random numbers and relative public information (likes, hometown, ecc) to perform some targeted advertising.
My very simple python script is bound to the italian version of Facebook, but is essentially very easy, uses Selenium webdrivers to login in Facebook and then search for numbers in the serachbox, looking for changes in the search suggestions to identify a match.
Since I don’t need people to get my Facebook friendship only because I’m in their phonebook, I changed the option value from everyone to friends.
I’ve created a little Genetic Algorithm library called JSGenetic, which takes care of the population management of a genetic algorithm.
As a test for this library, I’ve re-implemented an old project, the Genetic Environment, which is basically a simulation of an environment made by resources (water and food) which spontaneously grows or rains on the field, and by “animals”, which have a genetic code that specifies their behavior given what they see of the environment: they can actually “see” only north, south, east and west. The animals also have a little memory of 4 bits, and can understand if they are running out of water or food.
The genetic code of the animals is initially set to random, and is progressively processed by the genetic library which selects the individuals which survive more time.
After some generations, it’s clear that the individuals live longer because of the selection made on their genetic code, which defines how they behave.
The code of the JSGenetic library is on GitHub, and the live example of the code is on my github page at this address.
This is a little layout I developed on the concept suggested by a friend of mine, to whom is dedicated the project. The code is obscene, and the link is here.
Today I tried to implement a simple webapp which retrieves some RSS feeds from a given URL and then looks for the content and uses a hierarchial clustering classifier to cluster them in some kind of categories by content similarity.
Actually the implementation is really poor, and I’m not even sure it works. Anyway, it’s been long since I wanted try some kind of text classifier, and here we are. It extracts the text from the RSS feed, then indexes the words inside it and tries to use them as features for the algorithm. Short posts generally means bad results, especially without any kind of generalization (tokenization should be the word in this case) of the features. In fact, results are hardly understandable and I guess they’re random.
Anyway, here’s a list of what I (sort of) learned along the way:
That’s not so bad for a spare afternoon&evening.
As I said, I didn’t implement the algorithm myself, but I used a library from github, clusterfck.
Oh and I also used the jFeed jQuery plugin for parsing the RSS, but I slightly modified it to fetch the content of the entries and not to crash trying to detect IE.
Here‘s the link.