In machine learning, a model learns patterns from data during training, then uses those patterns to make predictions or generate outputs on new inputs. Developers evaluate the result on examples outside training. Using the trained model is called inference; it is different from continuing to train it.
Machine Learning: The Foundation of AI
Machine learning is one family of techniques within artificial intelligence (AI). It uses data to train a model for a task, rather than requiring someone to write every decision rule by hand. Learning from training data does not show that the model will perform well on every new input.
How Does Machine Learning Work?
In simple terms, machine learning works by employing algorithms to analyze data, learn from it, and make predictions or decisions based on that data. This machine learning can be broken down into three main steps:
- Data Collection: The first step is collecting a large amount of data, also known as a dataset. This dataset should contain examples of the problem you want the machine to learn about, such as images, text, or numerical data.
- Training: Next, the machine learning algorithm is exposed to the dataset. During this phase, the algorithm analyzes the data and learns patterns and relationships within it. The algorithm then uses this knowledge to create a mathematical model that can make predictions or decisions based on new data.
- Evaluation: Finally, the performance of the machine learning model is evaluated using new, unseen data. Once the model performs as expected, it can be deployed for real-world use. If not, the model may need further training or adjustments.
How Does Deep Learning Work?
Deep learning is a machine-learning approach that uses neural networks with multiple layers. The layers transform input data into representations useful for the task. The name describes a computational method; it does not establish that the system learns or understands as a person does.
In a deep learning model, data is passed through multiple layers of interconnected nodes called neurons. Each layer transforms the data, allowing the model to learn increasingly complex patterns and features. The final layer produces the output, such as a prediction or classification.
How Does AI Learn?
During training, learning methods adjust a model using data or feedback. After training, the model uses its learned parameters to produce outputs: inference. Some systems support further training, but a normal prediction or conversation does not by itself mean that the model’s parameters are being updated.
The methods described here use data to learn patterns that can be applied to new examples. The training process, the data, and the way results are tested all affect what the model can do.
When someone says that a machine has learned, ask what it learned to do and how that ability was tested. Those questions turn a broad claim into something you can assess.
The author generated this text in part with an OpenAI GPT large-scale language-generation model. Upon generating draft language, the author reviewed, edited, and revised the language to their own liking and takes ultimate responsibility for the content of this publication.
Google’s supervised-learning guide: training, evaluation, and inference
