機器學習Machine learning

• 07/14/2020
• • v
  • o

Machine learning 是 AI 或人工智慧的子集，可讓機器偵測模式並從資料中學習，而不需要明確 programed。Machine learning is a subset of AI, or artificial intelligence, which allows machines to detect patterns and learn from data without expressly being programed for it. Azure Machine Learning 解決方案可以提升您的計算見解。Azure Machine Learning solutions can advance your computing insights.

Azure 可為您提供最先進的機器學習功能。Azure empowers you with the most advanced machine learning capabilities. 使用 Azure Machine Learning，快速且輕鬆地建立、定型及部署機器學習模型。Quickly and easily build, train, and deploy your machine learning models by using Azure Machine Learning. Machine Learning 的 AI 可用於任何一種機器學習，從傳統到深度、監督及非監督式的學習。Machine Learning AI can be used for any kind of machine learning, from classical to deep, supervised, and unsupervised learning. 無論您想要撰寫 Python 或 R 程式碼，或使用零程式碼或低程式碼選項（例如 設計工具），都可以在 Machine Learning 工作區中建立、定型和追蹤高精確度的機器學習和深度學習模型。Whether you prefer to write Python or R code, or use zero-code or low-code options such as the designer, you can build, train, and track highly accurate machine learning and deep learning models in a Machine Learning workspace.

您甚至可以在本機電腦上開始訓練，然後向外延展到雲端。You can even start training on your local machine and then scale out to the cloud. 此服務也會與熱門深度學習交互操作，並增強式開放原始碼工具，例如 PyTorch、TensorFlow、scikit-learn 學習和光線和 RLlib。The service also interoperates with popular deep learning and reinforcement open-source tools such as PyTorch, TensorFlow, scikit-learn, and Ray and RLlib.

從閱讀 Machine Learning 的 解決方案開始，您可以在其中找到如何 設定第一個機器學習實驗的教學課程。Get started by reading Machine Learning solutions, where you'll find a tutorial on how to set up your first machine learning experiment. 若要深入瞭解機器學習的開放原始碼模型格式和執行時間，請參閱 ONNX runtime。To learn more about the open-source model format and runtime for machine learning, see ONNX Runtime.

機器學習解決方案的常見案例包括：Common scenarios for machine learning solutions include:

• 預測性維護Predictive maintenance
• 庫存管理Inventory management
• 詐騙偵測Fraud detection
• 需求預測Demand forecasting
• 智慧型建議Intelligent recommendations
• 銷售預測Sales forecasting

Machine Learning 檢查清單Machine Learning checklist

• 首先熟悉您自己的 Machine Learning，然後選擇要開始使用的體驗。Get started by first familiarizing yourself with Machine Learning, and then choose which experience to begin with. 您可以遵循下列步驟，使用 Jupyter 筆記本搭配 Python、視覺效果拖放體驗，或自動化機器學習 (AutoML) 。You can follow along with steps to use a Jupyter notebook with Python, the visual drag-and-drop experience, or automated machine learning (AutoML).

  • 瞭解 Machine LearningLearn about Machine Learning
  • 使用 Python 搭配 Jupyter 筆記本建立您的第一個機器學習實驗Create your first machine learning experiment with a Jupyter notebook with Python
  • 使用視覺效果拖放進行實驗Experiment with visual drag-and-drop
  • 使用 AutoML UIUse the AutoML UI
  • 設定您的開發環境Configure your dev environment

• 試驗更 高階的教學課程，以預測出租車費用、分類影像，並建立批次評分的管線。Experiment with more advanced tutorials to predict taxi fees, classify images, and build a pipeline for batch scoring.

  • 使用 AutoML 來預測出租車費用Use AutoML to predict taxi fees
  • 使用 scikit-learn 分類影像-學習Classify images with scikit-learn
  • 建立批次評分的 Machine Learning 管線Build a Machine Learning pipeline for batch scoring

• 遵循影片教學 課程，以深入瞭解 Machine Learning 的優點，例如無程式碼模型建立、機器學習作業 (MLOps) 、ONNX 執行時間、模型可解譯性和透明度等等。Follow along with video tutorials to learn more about the benefits of Machine Learning, such as no-code model building, machine learning operations (MLOps), ONNX Runtime, model interpretability and transparency, and more.

  • Machine Learning 的新功能What's new with Machine Learning
  • 使用 AutoML 建立模型Use AutoML to build models
  • 使用 Machine Learning 設計工具建立零程式碼模型Build zero-code models with Machine Learning designer
  • 管理端對端生命週期的 MLOpsMLOps for managing the end-to-end lifecycle
  • 將 ONNX 執行時間併入您的模型Incorporating ONNX Runtime into your models
  • 模型可解譯性和透明度Model interpretability and transparency
  • 使用 R 建立模型Building models with R

• 查看 AI 機器學習解決方案的參考架構。Review reference architectures for AI machine learning solutions.

下一步Next steps

探索其他 AI 解決方案類別：Explore other AI solution categories:

• AI 應用程式和代理程式AI applications and agents
• 知識採礦Knowledge mining