Day 8 - Tensors

A tensor is a multi-dimensional array of data. Just like a scalar is a single number, a vector is a one-dimensional array of numbers, and a matrix is a two-dimensional array, a tensor extends this concept to higher dimensions. Tensors are used to represent data and parameters in machine learning models.

Types of Tensors 📏

1. Scalar (0-D Tensor) 🏆

   • Description: A single number.

   • Example: 7, 3.14

   • Usage: Represents constants or single data points in computations.

2. Vector (1-D Tensor) 📈

   • Description: A one-dimensional array of numbers.

   • Example: [1, 2, 3]

   • Usage: Used to represent data features or model parameters in a list.

3. Matrix (2-D Tensor) 📉

   • Description: A two-dimensional array of numbers.

   • Example: [[1, 2], [3, 4]]

   • Usage: Commonly used to represent datasets, transformation matrices, and more.

4. 3-D Tensor 📦

   • Description: A three-dimensional array of numbers.

   • Example: [[[1, 2], [3, 4]], [[5, 6], [7, 8]]]

   • Usage: Often used in image processing where each image has multiple color channels (RGB).

5. 4-D Tensor and Higher 📐

   • Description: An array with four or more dimensions.

   • Example: In deep learning, a 4-D tensor might represent a batch of images, where dimensions correspond to batch size, image height, image width, and number of channels.

   • Usage: Utilized in complex models, such as those handling batches of multi-dimensional data.

Why Tensors are Essential in Machine Learning? 🚀

• Data Representation 🗃️: Simplifies data processing and learning.

• Efficient Computations ⚙️: Enables fast, GPU-optimized operations.

• Model Building 🏗️: Defines model parameters and layers.

• Versatility 🌐: Handles diverse data types and complex models.