Artificial Intelligence (AI)
AI refers to the simulation of human intelligence in machines that are capable of performing tasks that typically require human intelligence, such as understanding natural language, recognizing patterns, solving problems, making decisions, and learning from experiences.
Three main categories of AI are:
- Artificial Narrow Intelligence (ANI): “weak” AI. Can complete a specific task like win a chess game or identify people in images (Like in GooglePhotos) or NLP or automated chatbots on websites.
- Artificial General Intelligence (AGI)
- Artificial Super Intelligence (ASI)
- AGI and ASI are “strong’ AI that incorporates human behaviors more prominently, such as the ability to interpret tone and emotion.
Machine Learning (ML)
Machine Learning is a subset of AI that focuses on the development of algorithms and models that enable computers to learn from past data and improve their performance over time. Instead of being explicitly programmed, a machine learning system uses data to recognize patterns, make predictions, or take actions. ML includes techniques like regression, classification, clustering, and neural networks.
Three broad categories of ML are:
- Supervised Learning
- Unsupervised Learning
- Reinforcement Learning
Deep Learning (DL)
DL can be considered as neural networks with a large number of parameters layers lying in one of the 4 fundamental network architectures:
- Unsupervised Pre-trained Networks
- Convolutional Neural Networks
- Recurrent Neural Networks
- Recursive Neural Networks
| ARTIFICIAL INTELLIGENCE | MACHINE LEARNING | DEEP LEARNING |
|---|---|---|
| The study/process that enables machines to mimic human behavior through use of algorithms. |
Statistical methods enabling machines to improve with experience by learning from past data | Study that makes use of Neural Networks (similar to neurons present in the human brain) to imitate functionality just like a human brain. |
| AI is the broader family consisting of ML and DL as it’s components. | ML is the subset of AI. | DL is the subset of ML. |
| Search Trees and complex math | K-Mean, Support Vector Machines, etc. | We break the complex functionalities into linear/lower dimension features by adding more layers |
| Increase chances of success and not accuracy. | Increase accuracy | Highest rank in terms of accuracy when it is trained with large amount of data. |
| E.g.: Sentiment-based news aggregation, Image analysis and caption generation, etc. |
E.g.: Virtual Personal Assistants: Siri, Alexa, Google, etc., Email Spam and Malware Filtering. |
E.g.: Sentiment-based news aggregation, Image analysis and caption generation, etc. |
| Rule-based, knowledge-based, or data-driven. | Trial and error, receiving feedback in the form of rewards or punishments. | Multiple layers of interconnected neurons that process data in a hierarchical manner, allowing them to learn increasingly complex representations of the data |
Neural Network
Neural networks, also called artificial neural networks (ANNs) or simulated neural networks (SNNs), are a subset of ML and are the backbone of DL algorithms. The “deep” in deep learning refers to the depth of layers in a neural network. A neural network of more than three layers, including the inputs and the output, can be considered a deep-learning algorithm.
Working
- They are called “neural” because they mimic how neurons in the brain signal one another.
- Neural networks are made up of node layers – an input layer, one or more hidden layers, and an output layer.
- Each node is an artificial neuron that connects to the next, and each has a weight and threshold value. When one node’s output is above the threshold value, that node is activated and sends its data to the network’s next layer. If it’s below the threshold, no data passes along.
- Applications: Financial Forcasting, Medical Diagnosis, Gaming, Fraud Detection, Robotics, Image/Speech recognition, NLP, etc.
Limitations of Neural Networks
- Black Box Problem: ANNs are often considered to be a “black box” because it can be difficult to understand how they arrived at a particular decision or conclusion. Makes it difficult to trust them right?
- Heavily depends on the quality and quantity of data available for training. If data is missing/biased/faulty, results will be wrong.
- Require a lot of computational resources to train and run. This can be a limitation for real-time processing or for systems/applications with limited computational resources.
- Limited to supervised learning: Most ANNs are limited to supervised learning, which means they need to have labelled data to learn from. This makes it difficult to use them in unsupervised or reinforcement learning scenarios.
Other Terms
Natural Language Processing (NLP)
Machine learning algorithms are used to process natural language and extract meaning from it. For example, a chatbot can understand and respond to user queries.
Reinforcement Learning
Machine learning algorithms learn from experience and optimize decisions. For example, a game-playing AI can be used to learn how to play a game by trial and error. Reinforcement learning can be used in conjunction with deep learning, but it is not necessarily the same.
Generative adversarial networks (GANs)
GANs are a type of AI algorithm that can be used to generate new data from existing data. They are used to generate realistic images, videos and audio.
Large language model (LLM)
Large language model (LLM) is a type of AI algorithm that uses deep learning techniques and massively large data sets to understand, summarize, generate and predict new content. The term generative AI are a type of generative AI that has been specifically architected to help generate text-based content. (Chat GPT by open AI, BERT (Bidirectional Encoder Representations from Transformers) by Google)