Artificial intelligence buzzwords

Artificial intelligence

Artificial intelligence (AI) is a man-made intelligence carried out by a software program. It has the ability to perform tasks a human can do such as holding a conversation, driving a car, writing a letter, translating languages, learn new skills, etc. With proper training, the computer can even perform the tasks better than a human.

Generative artificial intelligence

Generative artificial intelligence (AI), also known as generative AI or GenAI, is a type of AI that can create new content based on existing data. Example are asking an AI to chat with a customer, asking an AI to generate an image, asking an AI to write a book, etc.

Machine learning

Machine learning (ML) is a subset of artificial intelligence, it is a process to make a software program to learn and improve from data. For example, an AI software program for recognizing handwritten digits can improve the accuracy of reading handwritten digit by training it with tens of thousands labeled training data. Just like how a kid learns the digits, we will feed the software program with a image that contains a digit and tells it what digit it is. After feeding it enough labeled digit images, it will eventually be able to tell what digit is on an image that it never seen before.

Supervised learning

Supervised learning uses labeled data to train a software program model to predict specific outcomes. It is provided with “correct answers” in the training data, and learn to predict the correct answers as it gets trained with more and more training data. A handwritten digit recognization model is an example of supervised learning. Other examples includes identifying spam emails, credit card fraud detection, weather forecasting, heart disease prediction, etc.

Unsupervised learning

Unsupervised learning refers to a software model works on unlabeled data to discover hidden patterns and structures within the data on its own. Unsupervised learning includes data preprocessing, customer segmentation, recommendation engine, social network analysis, grouping weather patterns, etc.

Deep learning

Deep learning is an advanced neural network model which has multiple hidden layers to transform data into more abstract representations before reaching to the final output layer. For example, in an image recognition model, the raw input may be an image (represented as a tensor of pixels). The first representational layer may attempt to identify basic shapes such as lines and circles, the second layer may compose and encode arrangements of edges, the third layer may encode a nose and eyes, and the fourth layer may recognize that the image contains a face.

Neural networks

Neural networks are a type of machine learning algorithm that uses a network of interconnected nodes to process data, learn patterns, and enable tasks such as pattern recognition and decision-making. It’s called neural networks because it mimics the complex functions of the human brain.

Input layer

Input layer refers to the first column of nodes/neurons in a neural network. As the name suggests, it contains the input data. For an image recognization model, the input layer could be a list of pixel values of an image.

Hidden layer

Hidden layer refers to the layers between the input layer and the output layer. For example, in an image recognition model, the first hidden layer may contain data to identify basic shapes such as lines and circles, the second hidden layer may contain data to compose and encode arrangements of edges, the third layer may contain data to encode a nose and eyes, etc.

Output layer

Output layer refers to the final layer in a neural network model that tells us the answer we are looking for. For example, in a handwritten digit recognization model, the output layer will tell us what digit is on the image.

Activation function

Activation functions are mathematical functions in artificial neural networks that determine a neuron’s output based on its inputs and weights. Basically, the function determines which value to pass as output and what not to pass as output for the neurons in the hidden and output layers. Sigmoid, ReLU (Rectified Linear Unit), Tanh (Hyperbolic Tangent), and Softmax are some of the popular activation functions.

Loss function

A loss function is also called a lost function, it is a mathematical function that measures how well a machine learning model predicts a data set. It calculates the difference between the model’s output and the actual data values. For example, a model that predicts the temperature for tomorrow, the model predicts the temperature will be at 55 degrees in fahrenheit, if the actual temperature comes out to be 60 degrees, then the difference is 5 degrees. Having the model to give 10 predictions, then calculates the average squared difference between the model’s predictions and the actual values, this will give us the Mean Squared Error. In machine learning, the goal is to make the model to produce more accurate results, the smaller the mean squared error, the more accurate the model is.

Gradient descent

Gradient descent is an optimization algorithm used to train machine learning models by minimizing errors between predicted and actual results. It heavily relies on derivatives, specifically partial derivatives, to determine the direction of steepest descent when minimizing a cost function, it calculates the gradient (which is a vector of partial derivatives) at each step to update the parameters in the direction that will most effectively reduce the error.

Natural Language Processing (NLP)

Natural Language Processing (NLP) in artificial intelligence refers the ability for computer to understand and interpret human language, both written and spoken, essentially enabling computers to “read” and “comprehend” natural language like humans do.

Large language model (LLM)

A large language model (LLM) is a type of artificial intelligence (AI) that can generate and understand human language. LLMs are trained on large amounts of data, often billions of words, and use neural networks to learn language patterns and rules. They can perform many tasks such as generate text in response to a prompt of a sentence or paragraph, translate between languages, summarize text, answer questions, analyze sentiment, recognize speech, etc.

Computer vision

Computer vision is a field of artificial intelligence (AI) that allows computers to analyze visual data and understand objects and people in images and videos. It’s a type of AI that aims to replicate how humans see and understand the world around them.