Without knowing these Terms, ML/AI will always feel Futuristic/Alien.
In the world of AI/ML and AWS, there’s a sea of terms that often get tossed around.
If you’ve ever felt overwhelmed by them, don’t worry — you’re not alone!
This article will break down the most commonly used terms in AI/ML and AWS, giving you a clear, concise definition of each term, that will bring you one step closer to AI/ML/AWS world.
🔥AI: Its a process of imitating human intelligence in machines.
[problem-solving, reasoning, learning, and understanding language]
🔥ML: A subset of AI, ML refers to the ability of machines to learn from data without being explicitly programmed. It uses “Algorithms to find Patterns” or insights from large datasets.
🔥Deep Learning: A subset of ML, that deals with “algorithms” inspired by the structure/function of the “Human Brain”, called artificial neural networks. [image, speech recognition]
NOTE: When “Learning” comes in picture “Algorithm” comes into picture.
Like in above 2 case, ML & Deep Learning works on ‘Algorithm’
🔥Neural Networks: A set of “algorithms” that attempts to recognize underlying relationships, in a set of data through a process, that mimics the way the human brain operates.
Neural networks are the foundation of deep learning.
🔥Computer Vision: A field of AI that enables machines to interpret and make decisions based on visual data, such as images and videos.
[object detection, image recognition, and face detection]
🔥Natural Language Processing (NLP): A branch of AI that focuses on the interaction between computers and human language, enabling machines to understand, interpret, and generate human language.
[sentiment analysis, text translation, and chatbots.]
🔥Algorithm: A set of instructions or rules, designed to perform a task or solve a problem. In ML, algorithms are used to “train models from data”.
MOST IMPORTANT ONE 👇🏻
🔥Model: A ML MODEL, is a “mathematical representation” of a real-world process.
In ML, it is a function, learned from data, that makes predictions or classifications based on new input data.
A Model is created, by training an algorithm, using data, and the result of this training process is the model.
🔥Training: The process of teaching a machine learning model by feeding it data and allowing it to learn patterns.
🔥Inferencing: The last stage/process of now, using a trained model, to make predictions or decisions based on new data.
Bulk Inferencing: Prediction generated in BULK.DOne in Offline mode.
Real-Time Inferencing: Predicting in real time like self-driving car does.
Dont loose the FOCUS here, just few more terms and you will have a good awarness atleast.
🔥Bias: Bias in AI refers to systematic errors or prejudices that can be introduced in the model due to biased training data or flawed assumptions. This can lead to unfair or inaccurate predictions.
🔥Fairness: Ensuring that AI models make decisions without unjustly favoring or discriminating against specific groups, especially in sensitive areas like hiring, lending, or criminal justice.
🔥Fit: The ability of a model to perform well on the training data.
A “good fit” means the model has learned the underlying patterns, while “overfitting” or “underfitting” may indicate issues in model performance.
🔥Large Language Models (LLM): AI models designed to process and generate human language on a large scale, typically trained on massive datasets.
Used in NLP tasks like language translation, summarization, and conversation generation.
🔥Structured Data: Data that is organized in a specific format or schema, like tables, spreadsheets, or databases.
🔥Unstructured Data: Data that does not have a predefined format, such as text documents, videos, or audio files.
🔥AWS SageMaker: It is a fully managed service that provides developers and data scientists with the tools to quickly build, train, and deploy machine learning models at scale.
🔥AWS Bedrock: It is a managed service that allows you to easily build and scale generative AI applications by leveraging foundation models from leading AI companies.
Okay glad you came this far. I know there must be few confusions, so lets try to know the differences.
Similarities & Differences
🎯AI vs. ML:
AI is the broader concept that encompasses the idea of machines performing tasks that would typically require human intelligence (e.g., reasoning, decision-making).
ML is a “subset of AI” that specifically focuses on using data-driven algorithms to learn from and make predictions based on data.
🎯ML vs. Deep Learning:
Deep Learning is a specialized area within ML.
While traditional ML algorithms, might use “structured data” (like tables)
Deep Learning models are designed to, automatically learn hierarchical features from raw data, especially unstructured data like images and text.
Deep learning typically requires much larger datasets and computing power than traditional ML.
Now, dont loose the momentum, complete the last most important chunk of the blog, “THE MODEL”
🚀🚀🚀🚀🚀🚀🚀🚀🚀🚀
When we hear, “Train the Model”, “Model fine-tuning” etc etc
So How the MODEL Looks ?👀
A model itself isn’t a file you can open (like an Excel sheet).
Instead, it consists of:
The learned parameters (weights and biases): These are the “knowledge” the model has acquired during training.
In linear regression, these would be the coefficients.
In deep learning, these are the weights in the neural network layers.
The architecture: The structure of the model, which defines how the data flows through the model.
For example, a neural network has an architecture with multiple layers, each containing nodes (neurons) connected to each other.
Model as a Saved File:
Once a model is trained, it’s saved into a file. For example:
TensorFlow/Keras: .h5 file
PyTorch: .pth file
Scikit-learn: Pickle file (.pkl)
These files store all the necessary information about the model, such as:
Model architecture (if applicable)
Learned parameters (weights, biases)
Hyperparameters used during training (like learning rate, number of epochs)
Once saved, the model can be used to make predictions on new data without needing to retrain it.
Last part: How Does the Model Make Predictions?
Once trained and saved, the model is ready for inference (making predictions):
You load the trained model from the file.
You feed new, unseen data into the model (like an image of a cat or a house with specific features).
The model uses its learned parameters to output a prediction. For example:
🌾In the case of the house price model, you input the square footage and the number of bedrooms, and the model predicts the house price.
🌾In the case of the image classification model, you input an image, and the model classifies it as “cat” or “dog.”🐈🐶
Summary:
🚀A machine learning model isn’t an Excel file or a chunk of code; it’s a trained mathematical entity (typically stored as a file) that encapsulates learned knowledge.
🚀The model learns patterns from the data during training and is then able to predict outcomes when given new data.
🚀In simple ML models (e.g., linear regression), the model might look like a mathematical equation, while in more complex models (e.g., neural networks), it consists of learned weights and biases stored in specialized formats like .h5, .pth, or .pkl.
=============================
If you’re worried about AI replacing jobs or questioning your ability to keep up, remember that every expert was once a beginner.
Stay curious & motivated — keep learning & moving forward.
Keep Calm, Keep Aware, Keep the Chin and Thinking UP !! You will do it !!
If you want any personal suggestion or a one-to-one call with me, will be more then happy to have one🌿
Let’s connect on Linkedin for a Hi !!
Now, Take a deep breathe and Go Learn🌏
Top comments (0)