|
所在平台: Udemy |
课程主页: https://www.udemy.com/course/hands-on-transfer-learning-with-tensorflow-20/
课程评论：没有评论
课程名称：使用 TensorFlow 2.0 的实用迁移学习 课程概述：迁移学习是一种在新问题上使用预训练模型的方法，当前在深度学习领域非常流行，因为它能够以较少的数据训练深度神经网络。在迁移学习中，已经训练好的机器学习模型的知识被应用到一个不同但相关的问题上。其基本思路是利用在有大量标记训练数据的任务中学到的知识，来处理数据较少的新任务。学习过程中不是从零开始，而是借用解决相关任务所学到的模式。 本课程将教授如何实现迁移学习，以解决不同的机器学习问题，并重新利用预训练模型来训练其他模型。通过实践示例，您将了解迁移学习的应用，并掌握它在不同深度学习领域广泛使用的原因。您将实现迁移学习在卷积神经网络（CNN）和循环神经网络（RNN）中的实际应用，如图像分类、文本分类、情感分析等。课程中将演示如何训练模型，以及如何使用预训练模型来进一步训练类似的未经训练的模型，从而深入应用迁移学习，解决现实世界中的深度学习问题。 完成本课程后，您不仅能够构建机器学习模型，还将掌握使用 tf.keras、TensorFlow Hub 和 TensorFlow Lite 工具进行迁移学习。 关于作者：玛格丽特·梅纳德-里德是机器学习领域的谷歌开发者专家（GDE），是开源机器学习框架 TensorFlow 的贡献者，并且是官方 TensorFlow 博客的作者。她撰写教程并在会议上发表关于设备端 ML、深度学习、计算机视觉、TensorFlow 和 Android 的演讲。玛格丽特领导着谷歌开发者小组（GDG）西雅图和西雅图数据/分析/机器学习小组，热衷于帮助他人入门人工智能/机器学习。她曾在华盛顿大学的专业及继续教育项目中授课，数年来一直致力于 TensorFlow 的研究并为 TensorFlow 2.0 的成功作出了贡献。
Transfer learning involves using a pre-trained model on a new problem. It is currently very popular in the field of Deep Learning because it enables you to train Deep Neural Networks with comparatively little data. In Transfer learning, knowledge of an already trained Machine Learning model is applied to a different but related problem.The general idea is to use knowledge, which a model has learned from a task where a lot of labeled training data is available, in a new task where we don't have a lot of data. Instead of starting the learning process from scratch, you start from patterns that have been learned by solving a related task.In this course, learn how to implement transfer learning to solve a different set of machine learning problems by reusing pre-trained models to train other models. Hands-on examples with transfer learning will get you started, and allow you to master how and why it is extensively used in different deep learning domains.You will implement practical use cases of transfer learning in CNN and RNN such as using image classifiers, text classification, sentimental analysis, and much more. You'll be shown how to train models and how a pre-trained model is used to train similar untrained models in order to apply the transfer learning process even further. Allowing you to implement advanced use cases and learn how transfer learning is gaining momentum when it comes to solving real-world problems in deep learning.By the end of this course, you will not only be able to build machine learning models, but have mastered transferring with tf.keras, TensorFlow Hub, and TensorFlow Lite tools.About the AuthorMargaret Maynard-Reid is a Google Developer Expert (GDE) for Machine Learning, contributor to the open-source ML framework TensorFlow and an author of the official TensorFlow blog. She writes tutorials and speaks at conferences about on-device ML, deep learning, computer vision, TensorFlow, and Android.Margaret leads the Google Developer Group (GDG) Seattle and Seattle Data/Analytics/ML and is passionate about helping others get started with AI/ML. She has taught in the University of Washington Professional and Continuing Education program. For several years, she has been working with TensorFlow, and has contributed to the success of TensorFlow 2.0 by testing and organizing the Global Docs Sprint project.
