DEV Community: Fardeen9065

Neural Network

Fardeen9065 — Sat, 07 Jan 2023 14:38:41 +0000

In deeplearning, neural network most of the time uses various layers for doing a work. The neural networks can be both simple and complex. The concept of working with several layers comes from the functions of the brain.

In this image, we are seeing a neural network with 3 layers. The input x is taken as 0th layer. Then the layer containing three circles is the 1st layer. The circles are usually called units and they are actually information. After that, the 2nd layer that comes is the output layer where we get the output after passing an input x to a machine learning model. Lastly we use logistic regression to find out if the data is effective or not. Now, we are also going to see an example of complex neural network layers. Below is the image of a more complex neural network layer.

In this image, we can see that there are more layers than the previous image. In this image, the outputs of a particular layer are working as the input of the next layer. In this case a formula is also working. For determing the result of a particular unit of a given layer, a formula is used which is the equation of a straight line.

Here, g = wa + b is the formula. In this formula w and a are vector quantities. This is all about neural network layers.

Array Vs List

Fardeen9065 — Wed, 04 Jan 2023 02:27:16 +0000

Arrays and Lists are similar data structures. But there are some differences between arrays and lists. Lists are dynamic data structures. It means that lists can be manipulated. In other words, lists can be changed. An element can be easily added or removed in a list. For example if we want to add an element 3 in the list [7,5,1,8]. In case of list, we can easily do it using the list.append() function. But in this case the length of the list changes. Before the length of the list was 4. After the addition of 3 the length of the list now has become 5 and the list is now [7,5,1,8,3]. But this method will not work for arrays. That is because the size of the arrays are fixed. It means if we take the same array [7,5,1,8] like before then we cannot just add 3 to it. The reason is if we add 3 to the list then the size of the list will be changed. What we can do is that we can create another array setting the length of the array 1 greater than the previous array. As the length of the array is 4 we have to make another array with the length of 5. Then we will copy the previous elements in the new array. After that at the last index we can add 3. Thus, these are the differences between arrays and lists.

Supervised and Unsupervised Learning

Fardeen9065 — Sun, 18 Sep 2022 03:36:20 +0000

Supervised and Unsupervised Learning are an important part of Machine Learning. These two models of machine learning are used frequently in many works.

So what is Supervised Learning? Supervised Learning is that part of machine learning which takes an input x and gives a corresponding output. We can take the example of house price predictor. For predicting the price of the house, we can take the size of the house as input x and predict the price of the house. But how will we do this? First, let's see the graph of size and price of the house.

In this graph we have plotted some points for house price. Now we can pass a straight line through these points and this straight is the prediction of the house of the house like this image below.

This method of passing a straight line for predicting an output is called regression. There is another part of Supervised Learning which is called classification. It has a fixed set of outputs. We can take the example of spam detection for an email. In this case, we can have only two outputs which are that the email will be a spam or it won't.

Now we can talk about Unsupervised Learning. Unlike Supervised Learning, Unsupervised Learning is used to find a pattern within a data and seperating a boundary for the data. We an see the example in the following graph.

Clustering is a good example of Unsupervised Learning.
In conclusion, Supervised Learning Learning takes an input variable x and finds the corresponding output y and Unsupervised Learning finds a pattern within the data and predicts a solution for the data.

Time Complexity

Fardeen9065 — Wed, 14 Sep 2022 17:20:49 +0000

Time complexity is an important part of algorithms in programming. By using time complexity, we can estimate how much time is needed for a code to run. Usually 3 symbols are used to denote the time complexity. They are big O notation, omega and theta. omega represents best case scenario, theta represents average case scenario and big O notation represents worst case scenario. Mostly during analyzing time complexity, worst case scenarios are calculated. That's why big O notation is used. There are some notations for time complexity like O(1), O(n), O(n^2), O(logn), O(nlogn) etc.

Firstly, we will talk about O(1). This actually means that our code will run for constant time. For example, we can write the python code likex = 9. This line of code will have a time complexity of O(1) which is constant. Similarly, if we write the code

x = 9
y = 5

what will be the time complexity here? In this code, each line has a time complexity of O(1).As each line of this python code has a time complexity of O(1) so will the time complexity of the code be O(1+1) which is O(2)?No, it will still be O(1) that is because O(2) is also a constant time and O(1) is used to represent constant time.

But what if we have a loop in our code? Let's write a code for that.

for i in range(10):
   print(i)

Here the print statement has a time complexity of O(1). In this for loop the print statement will run 10 times.So, the time complexity will be O(10) as the loop will run 10 times. But what if the loop runs n times? Then the time complexity will be O(n). So, for a single loop the time complexity will be O(n). But what if there are 2 loops in our code. For example:

for i in range(n):
  print(i)
for j in range(n):
  print(j)

Here, the time complexity will be O(n+n) which is O(2n). But for writing time complexity we ignore the constants. So the time complexity for this code will also be O(n).

Besides, we will talk about another time complexity which is O(n^2). This time complexity is for nested loops. For example:

for i in range(n):
  for j in range(n):
     print(i,j)

In this code as there is a nested for loop the code will run O(n*n) times which is O(n^2) times. In this case also if there is the constant then we will remove it.What will be the time complexity for this code?

for i in range(2n):
   for j in range(n):
       print(i,j)
for a in range(n):
  print(a)

In this code, will the time complexity be O(2*(n^2)+n)? No, still the time complexity of the code will be O(n^2). The reason is that we ignore constants in writing big O notation. So, 2 will not be written in big O notation. But another question arises which is why won't the time complexity be O(n)? The reason is we always take the highest power of the equation we get in big O notation.That's why the time time complexity of the code is O(n^2).