|
|
| Line 1: |
Line 1: |
| {{see also|Machine learning terms}} | | {{see also|Machine learning terms}} |
| ==Introduction== | | ==Introduction== |
| In [[machine learning]], [[inference]] is when you make [[prediction]]s, decisions or [[generate content]] by applying a [[trained model]] to [[unlabeled example]]s or [[prompts]]. | | In machine learning, [[inference]] refers to the process of using a [[trained model]] to make [[prediction]]s or decisions about new [[data]]. The trained model takes in [[input data]] and produces [[output]] predictions based on its patterns learned from [[training data]]. Inference is essential for making a [[machine learning model]] into a practical [[application]] as it enables the model to be utilized for its intended purposes such as [[classifying images]], [[creating text]], or [[making recommendations]]. |
|
| |
|
| ==Types of Inference in Machine Learning==
| | Inference can be performed in real-time, where predictions are made as new data becomes available, or batch mode, where predictions are made for a large set of data all at once. Speed and accuracy in inference are crucial factors when applying machine learning models since they directly impact their usability and usefulness in practical applications. |
| In machine learning, there are two primary forms of inference: supervised and unsupervised.
| |
| | |
| Supervised inference is employed when the training data used to train a model includes labeled examples, meaning that each desired output can be predicted based on its input features. Common applications of supervised inference include image classification, speech recognition and natural language processing.
| |
| | |
| Unsupervised inference, on the other hand, occurs when data used to train a model does not include labeled examples. In this scenario, the model learns to detect patterns or relationships in input data without being explicitly told what its desired output should be. Common applications of unsupervised inference include clustering, dimensionality reduction and anomaly detection.
| |
| | |
| ==The Inference Process in Machine Learning==
| |
| Machine learning typically involves several steps for inference:
| |
| | |
| 1. Data Preparation: In this step, input data is cleaned, transformed and organized so it can be fed into the model.
| |
| | |
| 2. Model Selection: This step involves selecting an appropriate model based on the inference task, input data and desired outputs.
| |
| | |
| 3. Model Training: In this step, the model is trained using prepared input data through optimization to adjust its parameters so that its predictions are as precise as possible.
| |
| | |
| 4. Model Evaluation: Finally, the trained model is evaluated against a set of test data to assess its accuracy and identify any potential issues.
| |
| | |
| 5. Model Deployment: Finally, the trained model is deployed into a production environment so it can be utilized to make predictions on new data that has yet to be collected.
| |
|
| |
|
| ==Explain Like I'm 5 (ELI5)== | | ==Explain Like I'm 5 (ELI5)== |
