I’ll be cross-posting material on both sites for some time but after a little while I’ll simply move away from this site to my own hosted site. So, if you are reading this in feeds or check up on the site regularly, you should bookmark the new address.

Just to recall: Regression comes in when you want to estimate a response variable () given explanatory variables (). The goal for linear regression is to come up with values for parameters () such that you get a ‘best possible fit’ for each jth instance in the dataset of m instances:

In matrix notation, that comes simply to:

(For ease, we assume that is a vector of all 1s thus making the y-intercept).

The idea of gradient descent is to come up with the best (locally) values of . It does this by repeatedly updating the value of using the formula:

If you have no idea what this is or if you want to know this in-depth, read till the end.

Here’s the Matlab code for this whole procedure I wrote at first. Note the use of W for :

% Gradient descent algo for linear regression
% author: Nauman (recluze@gmail.com)

%set the data
X=[1 1 1 1 1 1 1; 22 49 80 26 40 54 91];
Y=[20 24 42 22 23 26 55];
hold on;
plot(X(2,:),Y, 'x');

% set the actual values of W
W = [5.775 0.474]';
YAct = (W' * X);

% GRADIENT DESCENT
W = zeros(2,1);     % initialize W to all zeros
m = size(X,2);      % number of instances
n = size(W,1);      % number of parameters
alpha = 0.000025;    % gradient descent step size

disp('Starting Weights:');
W

% 1) do the loop for W estimation using gradient descent ------------
for iter = 1 : 20 % let's just do 5 iterations for now ... not til convergence
% for each iteration
for i = 1 : n       % looping for each parameter W(i)
    % find the derivative of J(W) for all instances
    sum_j = 0;
    for j = 1 : m
        hW_j = 0;
        for inner_i = 1 : n
            hW_j = hW_j + W(inner_i) * X(inner_i,j);
        end
        sum_j = sum_j + ( (hW_j - Y(j)) * X(i,j) ) ;
    end
    % calculate new value of W(i)
    W(i) = W(i) - alpha * sum_j;
end
% plot the thingy
newY = (W' * X);
gColor = 0.05 * iter;
plot(X(2,:),newY, 'Color', [0 gColor 0]);

% end 1) ------------
end
% output the final weights
disp ('Final calculated weights');
W
% output actual weights from formula in Red W = [5.775 0.474]'
plot(X(2,:),YAct, 'Color',[1 0 0]);
% finish off
hold off;

Then I figured that I was just doing loops and didn’t really need the matrices. So, I went back and thought about the whole nested loop thing and replaced the loop portion with this:

for i = 1 : n       % looping for each parameter W(i)
    % calculate new value of W(i)
    W(i) = W(i) - alpha * sum( ((W'*X) - Y) .* X(i,:) );
end

Seems suspiciously like the original formula, doesn’t it? Here’s what the plot look like. The red line shows the linear regression line calculated from a closed-form formula and the other lines show estimations of with the darkest colored ones being the initial guess and lighter ones are successive guesses.

And finally, as promised, here’s where you can learn all about this stuff: machine learning video lectures from Stanford.

Now, this isn’t strictly a newbie tutorial. It isn’t even a tutorial per se. It’s more like a guideline that will tell you where to look to find what you need. The reason is that if I write something specific, it’ll go out-of-date in a giffy. So, I’ll tell you the process of how I figured out the steps and you can (probably) reproduce them even if the specifics have changed. This might not be the best way to do it but it certainly gets you going. I had to take quite a few detours to get to these steps; so they might save you some time. Ok, let’s go.

First, I read up on what an LSM is. (If you’re reading this, you probably know what it is.) Then, I came across the tlj link I mentioned earlier and read what register_security function, security_operations struct and others do.

Then, I went ahead and created a directory in the security folder of the linux kernel source. (I used linux-2.6.33.3, so the I’ll assume the source is sitting in /usr/src/linux-2.6.33.3) Assume the name of the LSM is blabbermouth (since it does nothing but printk the hooks).

To tell the kernel build about your LSM, I needed to do three things: (1) Create a Makefile. That’s simple. See this archive for the files and patches. (2) Create a Kconfig file. Again, fairly straight forward for a skeleton LSM. (3) Tell the security Makefile and Kconfig files about my LSM i.e. Edit security/Makefile and security/Kconfig. (See the two patch files in the archive to figure out where to insert the new lines.)

Once that is out of the way, I needed the actual source of the LSM. So, I created the blabbermouth.c fie. Now, the first thing was to create an instance of the security_ops structure. This is complicated because there are several fields in this and if you don’t have a good IDE, this is going to be tough. Of course, you have the kernel source so you can simply “borrow” code from elsewhere. I went to security/selinux/hooks.c and copied me a nice tidy security_operations struct (you can find it right at the bottom). Of course, it points to different functions in the same file so I got those too. By the way, you can find the declaration of the security_operations structure in include/linux/security.h

Changed all the selinux_* function names to blabbermouth_* function names and replaced the bodies of all these functions with a return 0; Hence the name “skeleton LSM”. I also inserted some printk statements just so that I’d know that the thing was working.

Of course, when you’re working with the kernel, you have to make sure you cater to the configurations. So, I surrounded the relevant portions with the #ifdef CONFIG_SECURITY_BLABBERMOUTH and #endif to make sure it only gets built if configured as such.

Once that is out of the way, I created a new initialization function with void __init blabbermouth_init and told the module to call it on initialization through module_init macro. Also created an __exit blabbermouth_exit function and registered it through module_exit.

Phew. That was fun. Now, back to make menuconfig and selected the blabbermouth module from the security page.

Make, make modules_install and make install and I was good to go. Felt good to see the blabbermouth outputs in dmesg.

The most popular argument in favour of facebook is this: “boycotting facebook for the reason that it refuses to take action against offensive groups is like boycotting telephone or internet, because facebook is just a medium”. It’s a red herring because facebook is a moderated medium while telephone and internet are not. The admins at facebook have control over the content that gets posted (and they do exercise this control in many instances). If facebook didn’t have the power to remove the offensive pages or if it wasn’t possible for them to moderate them, they’d be home-free but neither of these things is true. So, facebook is not off-the-hook. A better analogy would be that of a newspaper and you should certainly boycott a newspaper if it publishes offensive material. Facebook chooses not to take offensive groups off and for that, they are to blame.

Another argument is that boycotting facebook would make no difference. This is complete nonsense. Every business needs customers and every customer counts. Even if I can get no one else to leave facebook, just the fact that I’m leaving will make a difference to facebook. Besides, muslims are a huge portion of the world’s population and we can always make a difference. It’s really quite easy. Once we boycott one product/service for being offensive to our religion, many alternatives will surface and they would be careful. So, we’ll have what we want and still use our favourite products/services. It just needs a little bit of time (which unfortunately, we’re not willing to sacrifice).

I will not get into a debate of “free speech” versus “hate speech”. It’s fairly obvious to those who think with an open mind that insulting someone’s deep-felt beliefs is not free speech. Regardless of who the target is, insulting their religious beliefs is hate speech. This applies to all religions and no one should be allowed to insult any one else’s religious icons.

A while back, when a Danish newspaper published offensive depictions of our Holy Prophet (sallallaho-alaihe wasallam), I posted the fatwa by Mufti Ibrahim Desai on my blog. You can see it here. I will re-iterate a few very important points here. They are applicable now as they were then:

1. Express anger and disgust within the framework of the law of the country. Use the law of the country to its full capacity.
2. Boycott all Danish products and products of anyone else who support Denmark in this issue or even sympathise with Denmark.
3. Do not confine expression of anger to the cartoon issue. Use the opportunity to express the disgust on Islamophobia in general and highlight some pertinent issues.
4. Do not conduct ourselves with our fellow non-Muslim citizens in a way that offend them and look for an opportunity to get back at us. We will then be the cause of them showing disrespect to Islam and Islamic values. Always express good character and conduct to all.
5. The most important issue in expressing our love and honour for Rasulullah [sallallaahu alayhi wasallam] is following his Sunnah. Expressing our anger at the disrespect to Rasulullah [sallallaahu alayhi wasallam] is important but it will be hypocritical if we do not emulate the outward appearance, beard, dressing, etc. and noble character of Rasulullah [sallallaahu alayhi wasallam]. Every one should obtain a kitaab on the Sunnats of Rasulullah [sallallaahu alayhi wasallam] and implement that in our lives and pass this over to others.

So, these are the reasons I’m leaving facebook. I’m not de-activating my account. I’m deleting it. I set the date to May 19. I will switch to another service and if that service does the same thing, I’ll move on again. I can live without social networks if I have to.

I urge you all to spread this message.

For the time being though, enjoy the screenshot.

Update: fixed the plugin with a new ‘View’. It now operates much better with a separate view for the output and controls. Also added is a ‘Hierarchy’ view for viewing the policy in a nice tree structure.

First, create a new project of type ‘plugin development’. If you don’t know what that means, try finding a basic tutorial for using eclipse. The following screenshot shows the options you need to set for the plugin project. (Click to enlarge.)

And

When the project is created, you’ll see the manifest file in the main editor view. (If you don’t, just click on it in the manifest project in the project explorer.) See the different tabs in the bottom of the view.

In the ‘Extensions’ tab, you’ll see the ‘All Extensions’ pane. Click on ‘Add’ and search for ‘actionSets’. You need to create an extension of type ‘org.eclipse.ui.actionSets’. This basically allows the plugin to extend the basic eclipse interface (i.e. a menu in our case).

Right-click on the actionSet just created and create a new ‘menu’. (You would probably want to change the properties, especially the name, of the actionSet in an actual project.)

Then, under the menu, create a new separator and give it a name. This will allow the placement of a menuitem later on.

Create an ‘action’ type under the actionset similar to the menu item. This will be the actual item that we can click on in the menu. Once you do that, give it a proper name and click on ‘class’ in the properties pane. This will allow you to associate a class with this action item. (This should remind you of action listeners but this is a lot more dependent on eclipse plugin architecture.) Give the class a name and place it in your package.

Now, finally, for some code. Open the class you just created and you’ll see that there is some placeholder text inserted in. (Not going into the theory of what this actually is and how it works. Code first, think later!) Here’s the code. I’ve formatted the additions you need to make in a bold, blue typeface here.


public class HelloWorldAction implements IWorkbenchWindowActionDelegate {
IWorkbenchWindow window;
@Override
public void init(IWorkbenchWindow window) {
// TODO Auto-generated method stub
this.window = window;

}

}

And running the plugin is easy. Right click the project, Run-As and ‘Eclipse Application’.

Finally, to deploy the project, just go to ‘File’ -> ‘Export’ and select ‘Deployable Plugins and Fragments’. Just give it an archive name and you will have a zip file that can be extracted in any eclipse directory to incorporate the new plugin in that project. You can also export it to a repository but that’s another story. Last, but not least, see the screenshot and marvel at what you’ve done.

What’s next? See this tutorial on how to create an RCP application in eclipse. Maybe someday I’ll write one of those too.

1. We can’t let it go. We can’t forget about it because it’s part of our Imaan to care about such things. We cannot tolerate someone who denies the finality of the prophethood of Hazrat Muhammad (Sallallah-o-alaih-e-wasallam). Period.
2. It doesn’t matter what other things he does. It’s not relevant any more.

So, if you’re a “Zaid Hamid Fan” here’s a simple flow chart that can help you decide what your opinion is regarding Zaid Hamid. Clear your mind and come clean with what you think about him. Then decide what the real situation here is. (Click the image to see full size.)

Confusion among Zaid Hamid Fans

I never write about modern cults because they’re usually very short-lived and anyone with the most basic of religious knowledge can understand the contradictions in their ideology. For example, I never bothered to condemn Javed Ahmad Ghamidi because his opinions are obviously deviant and anyone can figure out that he is either very incompetent or very dishonest. The case of Zaid Hamid is different because he never speaks openly of his religious beliefs.

Ever since I heard about Zahid Hamid, I had some lack of clarity about him. I browsed through his site and read the ‘frequently asked questions’ section but I wasn’t convinced. It probably had something to do with the fact that whenever someone starts his campaign condemning sectarianism, it’s usually because they have no ideology of their own or they have one that they are not willing to share.

There have been rumours (and denials) going on about his affiliation with Yousaf Kazzab. No body seemed to know what the truth was and I wasn’t prepared to comment until I knew for sure. Today, I looked at his video in which he explicitly addressed the question. They just decided to share this video (mostly because it was released by people who were strictly ordered not to release it) and unfortunately (for him), he finally messed up. These videos clearly show that he supports Yousuf Kazzab.

First, let me mention a few points here:

1. Regardless of what capabilities a person has or the (good) influence he has on the society, the religious soundness can only be judged by the beliefs that are depicted by the actions. Besides, no one who believes in a ‘false prophet’ can do good to the Muslim ummah.

2. Claims must be backed up by actions. If a person claims that he does not endorse the prophethood of anyone after Prophet Muhammad (Sallall-o-alaihe-wasallam) and yet affiliates himself with someone who declares himself to be a prophet, then that claim is worth nothing.

3. Now, about the video: If you see his videos about this subject (links following), you’ll know Zaid Hamid does not deny his affiliation with Yousuf Kazzab. He does say that he left him before the trial (which is not true) but does admit he was with him at start. This, in itself, would’ve been alright but once you are publicly known to be affiliated with someone, and that someone turns out to be a deviant and a liar, you have to openly deny any further association. Moreover, you must explicitly deny having the same beliefs as the deviant. This brings us to the next point.

4. He now clearly states that the judgement against Yousuf Kazzab was unjust and that the only reason he isn’t taking it up in court is that it will take ‘three years’ in high court. He wants a shariah court listening to this and that isn’t possible according to Pakistan’s penal code. Convenient. The problem, however, is that in making this claim, he has clearly admitted that he disagrees with the judgement against Yousuf Kazzab and thinks that he had done nothing wrong. I’ll get back to this precise point in a little while but notice that this is exactly what his opposers are saying. Even if he is not Yousuf Kazzab’s ‘khalifa’ (just assuming here), he endorses his beliefs and that is the core of the dispute at hand.

5. In his videos, he repeatedly claims that Maulana Abdus-Sattar Niazi disagreed with the death sentence for Yousuf Kazzab. He fails to mention Maulana’s later statement about a certain Zaid Zaman coming to him and misleading him about Yousuf Kazzab and his beliefs! Zaid (Zaman) Hamid himself has declared Maulana to be ‘one of the most respected scholars’. Surely, he can’t claim that Maulana was lying, can he? He tries to portray Yousuf Kazzab’s case as a Deobandi-barelvi problem because that’s an easy way out. Just blame it on sectarianism. Forgets to mention all those Barelvi ulema who opposed Yousuf Kazzab though.

6. Last, but not least, understand that Ulema-e-Haq do not go against any person out of spite. Almi-Majlis-e-Tahaffuz-e-Khatm-e-Nubuwwat (AMTKN) has always been home to the most prominent and world renouned scholars. Their piety and rigor in scholarly matters has never been challenged. I know this personally because I’ve read both sides of the argument. In cases past and present. Read the fatawa and see for yourself.

Finally, the links:

First of all, the fatwa from Jamia Binoria about Zaid Hamid. Read the fatwa here and the article ‘Rahbar k Roop mein Rahzan’ here. The article refers to a speech by Yousuf Kazzab and two of his disciples (including Zaid Zaman Hamid) which is the last one here.

Zaid Zaman Hamid clearly endorses the opinion that Yousuf Kazzab was wrongly accused. He even vows to drag into court Yousuf Kazzab’s opponents (more precisely AMTKN) “when shariah courts are established” in Pakistan.

Finally, take a look here for a few more scans regarding Zaid Hamid’s false claims.

So, here’s how I did it. This is more for archiving purposes than for teaching. So, use whatever you can. Post any queries and I’ll see if I can help.

First, though, you need to get the prerequisites installed on your system.  I had a lot of trouble with the compilation because xen build scripts don’t check for prereqs before starting. It kept stopping in the middle of the build because of missing packages. So, here’s the complete list of all the packages you need for building this version of xen. Just execute:

# sudo apt-get install build-essential mercurial gawk zlib1g-dev libzzip-dev libcurl4-openssl-dev xorg-dev gettext libncurses5-dev python2.5-dev texinfo texlive-latex-base texlive-latex-recommended texlive-fonts-extra texlive-fonts-recommended transfig bridge-utils uuid-dev


Xen also requires Dev86. Get that from here (You only need the binary, not the source. I got Dev86bin-0.16.17.tar.gz) Copy the .tar.gz file to / — your filesystem root and extract it.

Now, get the xen-unstable source. In my case, this was xen-4.0.0-rc3.

# hg clone http://xenbits.xensource.com/xen-unstable.hg

That might take just a little while. This only gets the hypervisor source. The dom0 kernel source will be downloaded by the build script. Start the building process:

# make world

In the future, if you wish to rebuild the code without losing all the configuration change, just make dist .

This builds the hypervisor, downloads the associated dom0 kernel source (which might take a lot of time), builds xen-tools and other smaller packages. Install the whole thing using:


# ./install.sh


After all that, you should have xen.gz and the kernel (which in my case was 2.6.31.6) in the /boot directory.

It’s a good idea to create an initrd image. To do so, execute the following two commands:


# depmod -a 2.6.31.6
# update-initramfs -c -k 2.6.31.6


You now need to add the following directives to /boot/grub/menu.lst to enable booting into xen:


title Xen 4.0 w/ Linux 2.6.31.6
root (hd0,1)
kernel /boot/xen.gz dom0_mem=524288
module /boot/vmlinuz-2.6.31.6 root=/dev/hda1 ro
module /boot/initrd.img-2.6.31.6


Where /dev/hda1 is your linux partition. (You can get that from sudo fdisk -l.)

Reboot and you should be in Xen. Start the Xen daemon and take a look at your brand new dom0.


# /etc/init.d/xend start
# xm list


Next tutorial on how to create a domU and then move on to more exciting stuff such as DRTM!

Pathetic, really.