AutoML:优势和限制
使用集合让一切井井有条
根据您的偏好保存内容并对其进行分类。
本部分探讨了 Cloud Storage 的一些优势和限制,
AutoML,
可帮助您确定 AutoML 是否适合您的项目。
优势
您可以选择使用 AutoML 的一些最常见原因包括:
- 节省时间:许多人选择使用 AutoML 来节省时间,是避免
大量手动实验来找到最佳模型。
- 提高机器学习模型质量:AutoML 工具可以全面搜索
以获得最高质量的模型。
- 构建机器学习模型时无需具备专业技能:附带效应
是实现机器学习的普及。用户无需
对机器学习算法或编程有深入了解,
开发机器学习模型。
- 对数据集进行冒烟测试:即使是专家级模型构建人员,AutoML
可快速为您提供基准估算,了解数据集是否有足够的
所有噪声。如果 AutoML 工具无法构建平平无奇的工具,
那么花时间构建一个良好的模型可能并不值得
手动更改。
- 评估数据集:许多 AutoML 系统不仅可以帮助您评估数据集,
也有助于评估输入到模型中的
训练这些模型的方法。即使您不使用生成的模型,AutoML
工具可以帮助您确定哪些特征可能不值得
并将其包含在最终数据集中。
- 为了落实最佳做法:自动化功能提供内置支持,方便您采纳
每次模型搜索的机器学习最佳实践
限制
但是,在选择 AutoML 之前,您应该考虑一些限制:
- 模型质量可能不如手动训练。使用通用化的
优化算法找到最佳架构组合,
超参数和特征工程,这通常意味着
积极主动、有充足时间的专家创建出质量更高的模型,
预测质量。
- 模型搜索和复杂性可能是不透明的。有了 AutoML,
深入了解该工具如何得出最佳模型。事实上,
输出模型本身可能由不同的架构组成,
超参数和特征工程策略。使用
AutoML 很难手动重现。
- 多次 AutoML 运行可能会显示更大的方差:由于
优化算法以迭代方式找出其认为的最佳值
不同的 AutoML 运行可能会搜索模型的不同部分,
并适度(甚至可能显著)结束
不同位置。使用 AutoML 进行重新调整以创建新的模型架构
与简单地重新训练上一个输出模型相比,差异可能更大
架构。
- 模型在训练期间无法自定义。如果您的用例需要
自定义或调整,AutoML 可能并不是
适合您的选择。
数据要求
无论您是使用自定义训练还是 AutoML 系统,您都可以
从零开始构建模型时需要考虑的一点是,
数据量。AutoML 的优势在于,您大多可以忽略
架构和超参数搜索并主要关注
数据。
还有一些专门的 AutoML 系统可以使用
会显著减少数据
迁移学习。对于
而无需使用数十万个样本来构建
一个图片分类模型,那么这些专门的 AutoML 系统只能使用
数百张已加标签的图片以及从现有图片中
图片分类框架模型。
AutoML 适合您的项目吗?
AutoML 可以帮助任何人(从新手到专家)使用机器学习来构建产品和
解决问题。如果您想确定 AutoML 是否适合您的项目,
请考虑以下权衡因素:
- 如果团队在机器学习构建方面经验有限,AutoML 是理想之选
希望提高工作效率并拥有丰富的经验的团队
则没有自定义要求
- 如果模型质量较高,选择自定义(手动)训练更为合适。
团队需要能够自定义他们的模型。在这些
手动训练可能需要更多时间进行实验和
构建解决方案,但团队通常可以实现质量更高的模型,
与 AutoML 解决方案相比。
如未另行说明,那么本页面中的内容已根据知识共享署名 4.0 许可获得了许可,并且代码示例已根据 Apache 2.0 许可获得了许可。有关详情,请参阅 Google 开发者网站政策。Java 是 Oracle 和/或其关联公司的注册商标。
最后更新时间 (UTC):2024-08-13。
[null,null,["最后更新时间 (UTC):2024-08-13。"],[[["\u003cp\u003eAutoML helps save time and improve the quality of machine learning models by automating the model selection and training process.\u003c/p\u003e\n"],["\u003cp\u003eAutoML democratizes ML by enabling users without specialized skills to build models and is useful for dataset evaluation and enforcing best practices.\u003c/p\u003e\n"],["\u003cp\u003eAutoML may have limitations such as potentially lower model quality compared to manual training, limited customization during training, and opacity in the model search process.\u003c/p\u003e\n"],["\u003cp\u003eLarge amounts of data are generally required for AutoML, although specialized systems utilizing transfer learning can reduce this need.\u003c/p\u003e\n"],["\u003cp\u003eAutoML is suitable for teams with limited ML experience or those seeking productivity gains without customization needs, while manual training is preferred when model quality and customization are paramount.\u003c/p\u003e\n"]]],[],null,["# AutoML: Benefits and limitations\n\nThis section discusses some of the benefits and limitations of\n[**AutoML**](/machine-learning/glossary#automl) and can\nhelp you decide if AutoML is right for your project.\n\nBenefits\n--------\n\nSome of the most common reasons you might choose to use AutoML include:\n\n- **To save time**: Many people choose AutoML to save time by avoiding extensive manual experimentation to find the best model.\n- **To improve quality of ML model**: AutoML tools can comprehensively search for the highest quality model.\n- **To build an ML model without needing specialized skills**: A side effect of automating ML tasks is that it democratizes ML. Users don't need to have a deep understanding of machine learning algorithms or programming to develop an ML model.\n- **To smoke test a dataset**: Even if you're an expert model builder, AutoML can quickly give you a baseline estimate for whether a dataset has enough signal in all of its noise. If the AutoML tool can't build even a mediocre model, it might not be worth spending your time trying to build a good model by hand.\n- **To evaluate a dataset**: Many AutoML systems help you evaluate not just the models they produce but also help evaluate the features that go into training those models. Even if you don't use the resulting model, AutoML tools may help you determine which of your features might not be worth gathering and including in your final dataset.\n- **To enforce best practices**: Automation includes built-in support to apply ML best practices to each model search\n\nLimitations\n-----------\n\nHowever, before choosing AutoML there are some limitations you should consider:\n\n- **Model quality may not be as good as manual training.** Using a generalized optimization algorithm to find the best combination of architecture, hyperparameters, and feature engineering for any random dataset often means that a motivated expert with enough time can create a model with better prediction quality.\n- **Model search and complexity can be opaque.** With AutoML, it is difficult to have insight into how the tool arrived at the best model. In fact, the output models themselves may be made from different architectures, hyperparameters, and feature engineering strategies. Models generated with AutoML are difficult to reproduce manually.\n- **Multiple AutoML runs may show more variance**: Because of how the optimization algorithm iteratively finds what it thinks are the best values for building the model, different AutoML runs may search different portions of the space and wind up in moderately (or even possibly significantly) different places. Retuning with AutoML to create a new model architecture may show more variance than simply retraining the previous output model architecture.\n- **Models can't be customized during training.** If your use case requires customization or tweaking during the training process, AutoML may not be the right choice for you.\n\nData requirements\n-----------------\n\nWhether you're using custom training or an AutoML system, one thing you can\ncount on when you are building a model from scratch is that you need large\namounts of data. The advantage with AutoML is that you can mostly ignore the\narchitecture and hyperparameters search and focus primarily on the quality of\nyour data.\n\nThere are also specialized AutoML systems that can train models with\nsignificantly less data because they use\n[**transfer learning**](/machine-learning/glossary#transfer-learning). For\nexample, instead of requiring hundreds of thousands of examples to build\nan image classification model, these specialized AutoML systems can use only\na few hundred labeled images along with transfer learning from an existing\nimage classification framework model.\n\nIs AutoML right for your project?\n---------------------------------\n\nAutoML can help anyone---from novices to experts---use ML to build products and\nsolve problems. If you are trying to decide if AutoML is right for your project,\nconsider these trade offs:\n\n- **AutoML** is a great choice for a team with limited experience building ML models or experienced teams that are looking for productivity gains and don't have customization requirements.\n- **Custom (manual) training** is more appropriate when model quality is important and the team needs to be able to customize their model. In these cases, manual training may require more time for experimentation and building a solution, but the team can often achieve a higher quality model than with an AutoML solution.\n\n| **Key terms:**\n|\n| - [AutoML](/machine-learning/glossary#automl)\n- [Transfer Learning](/machine-learning/glossary#transfer-learning) \n[Help Center](https://support.google.com/machinelearningeducation)"]]