「ML プロジェクトの管理」では、ML プロジェクトがアイデアから本番環境対応の実装に向けて進む過程で、その管理方法について説明します。このコースでは、ML 開発フェーズと、ML チームに一般的に見られる役割とスキルについて説明します。関係者と連携するための戦略について説明し、開発の各フェーズで ML プロジェクトを計画して管理する方法について詳しく説明します。
ML プロジェクトに固有の複雑さをわかりやすく説明することで、ML プロジェクトを管理するための強固な理論的なフレームワークを提供します。
このコースでは、従来の ML モデルに焦点を当てます。生成 AI に注目が集まっていますが、従来の ML は Google において重要な役割を果たしており、マップでの移動時間の予測、フライトでの航空券料金の見積もり、Google Cloud の TPU 割り当ての予測、YouTube での関連動画のおすすめなど、多くのサービスやプロジェクトの基盤となっています。
一般に、従来の ML プロジェクトの管理原則は、生成 AI プロジェクトの管理の場合と同じです。大きな違いがある場合 このコースでは 生成 AI に関連するアドバイスとガイダンスを提供します
前提条件:
ML の基本を理解している必要があります。ML のコンセプトの概要については、 ML の概要をご覧ください。ML の実践的な概要については、ML 集中講座をご覧ください。
[null,null,["最終更新日 2024-03-12 UTC。"],[[["\u003cp\u003eThis course provides a comprehensive framework for managing machine learning (ML) projects, guiding you through all stages from ideation to production.\u003c/p\u003e\n"],["\u003cp\u003eIt covers key aspects such as defining project phases, planning and management strategies, establishing success metrics, and implementing responsible AI practices.\u003c/p\u003e\n"],["\u003cp\u003eWhile focused on traditional ML models, the course also offers insights into managing generative AI projects, highlighting common principles and key differences.\u003c/p\u003e\n"],["\u003cp\u003eTo benefit from this course, you should have a basic understanding of machine learning and have already determined that ML is the appropriate solution for your problem.\u003c/p\u003e\n"],["\u003cp\u003eIt's estimated to take approximately 90 minutes to complete this course, equipping you with the necessary skills to effectively manage your ML projects.\u003c/p\u003e\n"]]],[],null,["# Managing ML projects\n\n*Managing ML Projects* shows you how to manage an ML project as it progresses\nfrom an idea to a production-ready implementation. The course covers the\nML development phases and the roles and skills\ntypically found on ML teams. It discusses strategies for working with\nstakeholders and provides details on how to plan and manage an ML project\nat each phase of development.\n\nBy demystifying the complexities inherent in ML projects, the course\nprovides a solid theoretical framework for managing ML projects.\n\nThe course focuses on traditional ML models. Although generative AI is in the\nspotlight, traditional ML plays a vital role at Google, underpinning many\nservices and projects, from predicting travel times in Maps to estimating the\nprice of airline tickets in Flights, from predicting compute quota for Google\nCloud customers to recommending relevant videos in YouTube.\n\nIn general, the principles for managing traditional ML projects are identical\nfor managing generative AI projects. When there's a significant difference, the\ncourse provides relevant generative AI advice and guidance.\n| **Estimated Course Length:** 90 minutes\n| **Objectives:**\n|\n| - Define the phases and elements of an ML project.\n| - Describe how to plan and manage an ML project.\n| - Determine business and model success metrics.\n| - Recognize the iterative process of running ML experiments.\n| - Design a solution for productionizing ML pipelines.\n| - Implement responsible ML and AI practices at each development phase.\n\n**Prerequisites:**\n\n- You should have a basic understanding of machine learning. For a brief introduction to machine learning concepts, see [Introduction to Machine Learning](/machine-learning/intro-to-ml). For a hands-on introduction to machine learning, see [Machine\n Learning Crash Course](/machine-learning/crash-course).\n- You should first verify that ML is the right approach for your problem. If you haven't framed your problem in terms of an ML solution, complete [Introduction to Machine\n Learning Problem Framing](/machine-learning/problem-framing).\n\n\u003cbr /\u003e"]]