Skip to main content

Machine Learning

 

 Machine Learning:

The field of study that gives computer the ability to learn without being explicitly programmed.

 Types:

  • Supervised Learning
  • Unsupervised Learning  
  • Recommender Systems
  • Reinforcement Learning   

Supervised Learning:

    It is a method in which we give our algorithm both inputs and outputs, and the algorithm automatically generates an algorithm or mappings, so that at some point, that generated algorithm is so effective that if we give them inputs, respective outputs are automatically generated. 

The Two Main Types

  1. Regression: Predicting a number (e.g., predicting the price of a house).

  2. Classification: Predicting a category (e.g., deciding if an email is "Spam" or "Not Spam"). 

 

     Example:

 

  Here in this example, to predict the prices of houses, we plotted a graph, and we observed that we could use the function of curve to find the unknown value. But computers can't plot the graph and analyze here function of curve would be better. So, we need to figure some way to decide which function would be better here, or maybe some general function which can be applied to anything (right now i am guessing, function of derivation would be applied).

 Regression: 

 This type is called regression. In this type, we are trying to predict the number from infinitely many possible outputs.

In simple terms, regression is about predicting a specific number.

If you're trying to figure out a price, a temperature, or a score based on other data, you're doing regression.

Classification:

In classification, we usually do identification, like identifying a spam or identifying whether patient has breast cancer or not? In it the outcomes are limited (finite), not like regression, in which possible outcomes are infinite.

Key Types

  • Binary Classification: Only two choices (Yes/No, 0/1, True/False).

  • Multiclass Classification: More than two choices (Red, Blue, or Green). 

    The model looks at the data features and tries to find a boundary that separates the different groups. When you give it new data, it just checks which side of the boundary that data falls on.

    Want to see the common algorithms used for this, or should we talk about the difference between this and Regression?

     

    Example:

      

     

    SUMMARY:

      

     

Unsupervised Learning: 

A type of learning, in which we just give inputs to the model and it automatically find some meaningful information like structures, groups or meaningful outliers (exceptions).

Types:

1. Clustering

  • Definition: Automatically grouping data points that are "similar" based on their features.

  • Logic: If points are mathematically close to each other, they become a group (cluster).

  • Example: Grouping news articles by topic (Sports, Tech, Politics) without being told what the topics are.

2. Anomaly Detection

  • Definition: Identifying data points that are significantly different from the "normal" pattern.

  • Logic: The model learns what "average" looks like and flags anything that sits way outside that range.

  • Example: A bank flagging a $5,000 transaction in another country when you usually spend $20 on lunch locally.

 

 

 

Labels:

Comments

Post a Comment

Popular posts from this blog

Java Servlets

  According to me its just a translator, who manages the HTTP requests and handle them to the appropriate function of the API. So then, API turns out to be a term or set of those functions which are kept public. Now let us see in little bit of detail what servlets are: For that we need to see few terms: Static Web Pages: Static web pages are the pages, which are already built, and whose content and design always remain the same.  For Example: About us page of some website. Contact info page of some website Dynamic Web Pages: Dynamic pages are not already defined. They are dependent on search queries. For Example: If we search top 5 books, and click on search button, result would be different. If we search top 5 countries and enter search, result would be different.  So, the search result for every search is different. Working: Now if we have to search static page let's say the first page of any website (which is same for every use in every country) then the process is sim...
Post a Comment
Read more

OnClick and {}

1. The onClick Mystery: Is it the same? Yes, the concept is 90% the same. If you know how to handle a click in plain HTML/JavaScript, you already know how to do it in React. The logic flow is identical: Event: A user does something (Click). Listener: The element hears it ( onClick ). Handler: A specific function runs to deal with it. The "React Twist" (The 10% difference): There are two tiny syntax differences you must memorize: CamelCase: In HTML, it is onclick (lowercase). In React, it is onClick (camelCase). Function, not String: HTML (Old): onclick="doSomething()" (A string of code). React (New): onClick={doSomething} (Passing the actual function itself). So yes, your intuition is correct: Whenever you need to handle logic (clicks, hovers, typing), you will stick to this pattern: on[Event]={functionName} . 2. The Curly Braces { } : The Portal This is the most important syntax rule in JSX. Imagine your code file has two "Modes": HTML Mode: W...
Post a Comment
Read more

Servlet Context and Servlet Config

  Problem without init-parameters? Imagine you write a servlet that connects to a database. If you write the database URL, username, and password directly in your Java code, you have a big problem: What if the database password changes? You have to find the code, change it, recompile the .java file into a .class file, and redeploy the entire application. What if you want to use the same servlet in a different project with a different database? You can't. The code is stuck with the old settings. This is where init parameters come in. They let you move this configuration data out of your code and into the deployment descriptor ( web.xml file). How init-parameters help when we use them in web.xml file? They help us in a way that we don't have to find that particular part of code to change it. And also, it saves us the hectic of changing source code and recompiling that again. Anyone with .war file can change the web.xml file  and use it with class file without having the ne...
Post a Comment
Read more