Tag: code

Saving trained model
In most cases, for training the model with the dataset we have is very time consuming and also processing hungry job which is costly task. To test in our development environment we have to do is test the trained model or use the trained model for in production without going for multiple training.

If you have done some ML project you would have understood, how time and processor consuming task it is even when done in GPUs. For a application to use the model and train each time the application runs is unacceptable, so we can save the current trained state of the model for later use without of retraining the model on the same dataset again and again.

We can accomplish this in python using some packages like
- Pickle (Python Object Serialization Library)
- Joblib (One of the Scikit-learn Method)
Pickle?

You might have heard this term somewhere when you go though ML articles or doing projects. This library is popular for Serialization(Pickling) and Marshalling (Unpickling). Pickling is the process of converting any Python object into a stream of bytes in hierarchy.Unpickling is process of converting the pickled stream of bytes to original python object following the object hierarchy.

Example:

Serialization (Pickling)
```
import pickle

pickle_file = 'string_list_pickle.pkl'
names = ['apple', 'ball', 'cat']

store_pickle = open(pickle_file, 'wb')
pickle.dump(names, store_pickle)
store_pickle.close()
```
(more…)
13th February 2017
Simple Neural Network for absolute beginners

This is my first post regarding artificial intelligence (AI). But I promise to include as much as I can from understanding Simple Neural Network (NN) to deep learning through little theoretical but lots of practical implementations. I will also include simple projects where possible. Lets begin building then.

I like Python a lot so most of the works will be done in Python. Later on I am hoping to develop them in Java and Scala.

To learn NN we will not be using any NN libraries but some mathematical libraries, ie. numpy.

Learn basics of Numpy HERE .

To begin building NN which is supposed to mimic how our brain works, we have to understand little bit of our own Brain.

An averaged sized Brain includes of 100 billion neurons connected by synapses. Neurons are the basic unit of brain which plays major role for all the tasks done by brain. Blah blah blah… its better you go through this well written article (A Basic Introduction To Neural Networks).

This tutorial we will be building a Artificial unit of this very Neuron. I consider you know matrices which we will be using as mathematical foundation for building NN with numpy.

Our simple ANN will include three inputs and a output. (Input: 3, Output: 1). This neuron we will build should classify a basic problem of classification. We will use various different training algorithms to train our neuron for classification.

So our neuron will have very small dataset for training (a deeplearning model will need very very large dataset for better performance) which will be enough in this problem.

Example# Input A Input B Input C Output Y
1 0 0 1 0
2 1 1 1 1
3 1 0 1 1
4 0 1 1 0
5 1 0 0 ??? (1 expected)

So what will be the output for the last data row (Row #5)?.
(more…)

29th January 2017
Bloom filter – Python

A Bloom Filter is a space-efficient probabilistic data structure, to test whether an element is a member of a set. This algorithm for data representation ensures the element is in the dataset or not. This may have false positive matches but not false negative result.

The most common use of Bloom Filter Algorithm is to see if an elements is in the disk before performing any operations. This reduces the I/O for lookups dramatically over large datasets.

Consider we have to check a email address to search from large dataset of emails of millions. Searching all the emails in memory definitely is very inefficient and takes lots of time. We can create bloom filter bit array, which is very small compared to the original dataset and has almost same required result. This is what used in Yahoo Mail. When you log into Yahoo mail, the browser page requests a bloom filter representing your contact list from Yahoo servers. The Bloom Filter is compact and easily fits in your browser cache. In Yahoo, email is verified if it is in existed in your contact list.

Another scenario, consider we have to get unique counts in the dataset, then we can use bloom filter to test if certain pattern or element has already been seen or not in the dataset. Of course this creates some false positives, but this might be much efficient than to compare everything from the memory.

Apache HBase uses bloom filter to boost read speed by filtering out unnecessary disk reads of HFile blocks which do not contain a particular row or column.

Quora implemented a sharded bloom filter in the feed backend to filter out stories that people have seen before. It is much faster and more memory efficient than previous solutions (Redis, Tokyo Cabinet and DB) and saves hundreds of ms on certain types of requests.(Tao Xu, Engineer at Quora)

Transactional Memory (TM) has recently applied Bloom filters to detect memory access conflicts among threads.(Mark Jeffrey, modeled Bloom filters for concurrency conflict detection).

Not to mention Facebook ( Typeahead Search Tech Talk (6/15/2010) ),, LinkedIn ( Cleo: the open source technology behind LinkedIn’s typeahead search | LinkedIn Engineering), Bit.ly (dablooms – an open source, scalable, counting bloom filter library ) have implemented their own Bloom Filter.

More examples ? Go here https://en.wikipedia.org/wiki/Bloom_filter#Examples

Ok then, enough of the usage of Bloom Filter. We will be developing our own for searching email address from a list in Python.
(more…)

7th January 2017
Singleton Design Pattern
Singleton pattern restricts the instantiation of a class and ensures that only one instance of the class exists.
```
public class Singleton {
	private static Singleton instance;
	private Singleton() {}
	public static Singleton getInstance() {
		if (instance == null) {
			instance = new Singleton();
		}
		return instance;
	}
}
```
The above code is the general class definition of Singleton Design Pattern.

The above code guarantees the single object creation of the class. The only way to retrieve an instance of the class is to call getInstance() method.
4th January 2017
How many drive partitions can we assign?
Generally in a computer system we can assign 4 primary partitions, or we can assign 3 primary partitions and 1 extended partition with as many logical partitions as you want but withing the range of A – Z and assign any drive letter to the drives in windows. Similarly in linux its from sda-sdz. In the deeper level, to MBR we can describe more detailed and quite understanding way for why only 4 partitions. A MBR stored in 512 bytes of code in first sector of storage device.

The above is the memory map of MBR. This shows how many bytes are allocated in the code and which part of the code does what specific task.

In this section our concern is about the partition table only. That is at address 01BE, which is of 64 bytes. The complete MBR code would look like below.

The highlighted code is the partition table code which is 64 bytes long. Each 16 bytes of code represents 4 partitions.
```
Partition 1:80 20 21 00 83 FE FF FF 00 08 00 00 00 F8 5F 02 
Partition 2:00 FE FF FF 05 FE FF FF FE 07 60 02 02 F0 1F 00 
Partition 3:00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
Partition 4:00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
```
In the first partition from above
80: The first byte represents the bootable or non-bootable status of the partition. If the value is 80 the partition is bootable else if value is 00 it is not.
20 21 00: (Cylinder-Head-Sector Address) The first byte is represents head, Second represents “almost the head” and the last byte represents the Cylinder
83: Represents Partition type in our case XX means
FE FF FF: Cylinder Head Sector Address of the last absolute sector in partition
00 08 00 00: Logical block addressing of first absolute sector in the partition
00 F8 5F 02 : number of sectors in partition in our case (sectors-in decimal)

You can simple do the following:
```
$file <EXTRACTED_MBR_FILE>
```
Doing this you will the extracted data with partition info displayed.

https://www.bydavy.com/2012/01/lets-decrypt-a-master-boot-record
14th October 2013