Fundamentals of AI-ML
Relation between AI | ML | Generative AI
AI or artificial intelligence is a field of study that encomapasses building or creating systems that can perform tasks that require human intellidence. Think of a self driving car - the autonomous car driving system is able to sense its environment and then makes decisions like a human e.g., on seeing a red traffic light apply brakes, based on the GPS direction make a turn at a specific street etc. Another example is a system in which the player is replaced by a machine that shows intelligence of a real human. These AI based gaming systems can not only play the game but most time win against humans
Neural networks
A neuron is a specialized cell in the human brain that is the fundamental building block of a brain
A neuron in the context of human brain is a specialized cell that is the fundamental building block of the brain. Human brain is composed of billions of neurons that are inter-connected. These neurons get activated in response to external stimuli e.g., When there is a sound, it activates the neurons directly connected to the nerve endings from the ear.
These neurons process the auditory input, transforming it into electrical signals. They then pass these signals to connected neurons. This process continues as each neuron processes the received information, generating its output and transmitting it to subsequent neurons. Eventually, this chain of information processing culminates in the neurons conected to the vocal chord system, generating a vocal response. A neuron in an AI system, exhibits a behavior that is similar to the neuron in human brain.
Deep Learning Vs. Generative AI models
| Feature | Deep Learning | Generative AI |
|---|---|---|
| Primary Goal | Generalize from historical data to make accurate predictions on new data. | Generate new, synthetic data similar to the training data. |
| Example Use Cases | Image recognition, recommendation systems, speech recognition. | Image generation, essay writing, music composition. |
| Training Data Requirements | 100s of thousands to millions of samples. | 100s of gigabytes to a few terabytes. |
| Training Time | Days to weeks. | Weeks to months. |
| Tasks | Trained for specific tasks. | Can perform multiple generic tasks out of the box. |