Save Big on Coursera Plus. 7,000+ courses at $160 off. Limited Time Only!
Neural networks can be configured in various ways depending on the type of data and objectives. This course will help you understand how to properly choose a neural network architecture for image or audio data.
Deep learning, as opposed to machine learning, allows a more robust way to deal with image and audio data for various data science problems such as classification and clustering. This is largely due to the power of neural networks and their ability to learn the proper features to represent images and audio data without having to handpick such features as one would in traditional machine learning. However, understanding the correct architecture of the neural network is integral to get the best possible result. In this course, Using Neural Networks for Image and Voice Data Analysis, you’ll gain the ability to transform image and audio data and represent its numerical form (feature vector) to be fed to a neural network. First, you’ll explore the way data scientists define problems in image recognition, object detection, and speech-to-text in terms of vectorization and the expected format for how neural networks will ingest and infer this data. Next, you’ll discover how to properly assess different neural network architectures, from the most rudimentary ones such as vanilla CNN, to the more advanced ones such as transformer-based models, to properly solve a particular image or audio-based problem, as well as their strengths and weaknesses. Finally, you’ll see how to use a deep learning framework called PyTorch to easily test out different implementations of neural networks, and see how it trains against the data as well as measures the model’s performance using various metrics. When you’re finished with this course, you’ll have the skills and knowledge of neural networks needed to properly discern and execute a neural network architecture given image and voice data.
Deep learning, as opposed to machine learning, allows a more robust way to deal with image and audio data for various data science problems such as classification and clustering. This is largely due to the power of neural networks and their ability to learn the proper features to represent images and audio data without having to handpick such features as one would in traditional machine learning. However, understanding the correct architecture of the neural network is integral to get the best possible result. In this course, Using Neural Networks for Image and Voice Data Analysis, you’ll gain the ability to transform image and audio data and represent its numerical form (feature vector) to be fed to a neural network. First, you’ll explore the way data scientists define problems in image recognition, object detection, and speech-to-text in terms of vectorization and the expected format for how neural networks will ingest and infer this data. Next, you’ll discover how to properly assess different neural network architectures, from the most rudimentary ones such as vanilla CNN, to the more advanced ones such as transformer-based models, to properly solve a particular image or audio-based problem, as well as their strengths and weaknesses. Finally, you’ll see how to use a deep learning framework called PyTorch to easily test out different implementations of neural networks, and see how it trains against the data as well as measures the model’s performance using various metrics. When you’re finished with this course, you’ll have the skills and knowledge of neural networks needed to properly discern and execute a neural network architecture given image and voice data.