📢 On the auspicious occasion of Dhanteras & Deepawali, I announced that my book : 📕"Objects, Data & AI" is out and open sourced at the following webpage: https://t.co/XU9M3J4OCO 🏹 Content of the book ranges from simple design patterns and use of databases to evolution of AI.

But what about data? Prompt Engineering will not save you! Data will but you must learn how to manage it! Read the article and understand the details of the game. https://t.co/EnaGpHTM9H

👷♂️ Software Architecture Series���— Part 36. ⚔️Working with Metrics and Loss Functions in AI/ML. ✌️ This article is only helpful if you are trying build a model for your custom use case rather than leveraging an already available LLM for your problem. https://t.co/lRJgrjaDG5

👷‍♂️Software Architecture Series — Part 35. Why ‘Attention’ in Transformers are Special?

To be truly independent, we must be connected with the Self, where we perceive the events of outside worlds as mere observations, which do not alter the state of Self, which is anyway eternal constant.

We quickly try to evaluate ourselves based on certain parameters. However, this very evaluation beats the purpose of independence.

We live in an interconnected world, where our subconscious is exposes to the interaction with outside world, be it visuals, sounds, texts, etc. Hence, the very notion of being independent in our mind comes from this exposure.

For being truly independent, the equation must be P(A⊓B) = P(A)xP(B). A sample of this would be the probability of tossing a coin resulting in head or tails is completely independent of the fact whether it will rain tonight.

What this means is if we know that event A has occurred then we surely know that we event B has not occurred, which makes the two events dependents on each other. Hence, the illusion of being disjoint in a large sample space does not hold up to being independent.

Since, the events are disjoint (not sharing anything), we can intuitively guess, P(A⊓B) = 0 However, probability of the events happening individually remains greater than zero, i.e. P(A) > 0 and P(B) > 0.

This basically means occurrence of one event provides no context or effect on occurrence of other event. Now consider a large sample space and imagine two disjoint events A and B within it.

Mathematical definition of Independence In a world where everyone is racing to be independent, let us dig mathematically what independence really means? Technically, independence is a relation about information.

Preview: https://t.co/D7pR7g7woS

📉To keep the costs low and yet deliver results, one needs to build their own models from scratch and get hold of their data. 📺I have created a live playbook that is intended to develop intuitions to build a neural network from scratch. Access it here:

🗑️LLMs are nothing but deception to indulge the general competition in mundane activities so that big tech's can keep pursuing their interest. 🤑No organization has been able to successfully make profit by using LLMs and they are not even meant for the same.

📔If you are a programmer and want to cover the basics of Machine Learning and AI to progress in your career, you can refer to my book:

📜In this article, we dive deeper to understand a widely accepted technique called Principal Component Analysis, which is commonly used at almost every big tech to reduce data to lower dimensions yet retaining the dominant patterns within.

Machine Learning or real life, we try to prune away unnecessary information and retain the important bits to focus on the bigger picture. Not all the features in a dataset are informative or useful.

⏮️In the last article, we covered how to capture dimensionality in real-world data. This step is crucial to make neural networks smart and adapt to our use-case. However, as datasets become more complex, it becomes increasingly difficult to analyse them using traditional methods.

👷‍♂️ Software Architecture Series — Part 34. Deep Dive into Principal Component Analysis(PCA). https://t.co/1RFYN9tIRf

💰Bonus: This article uses a recursive approach to help you understand how a neural network works in simpler terms.