Pure Language Processing in TensorFlow

About this Course

In case you are a software program developer who desires to construct scalable AI-powered algorithms, you want to perceive the best way to use the instruments to construct them. This Specialization will train you greatest practices for utilizing TensorFlow, a well-liked open-source framework for machine studying.

In Course 3 of the deeplearning.ai TensorFlow Specialization, you’ll construct pure language processing methods utilizing TensorFlow. You’ll study to course of textual content, together with tokenizing and representing sentences as vectors, in order that they are often enter to a neural community. You’ll additionally study to use RNNs, GRUs, and LSTMs in TensorFlow. Lastly, you’ll get to coach an LSTM on current textual content to create unique poetry! The Machine Studying course and Deep Studying Specialization from Andrew Ng train a very powerful and foundational rules of Machine Studying and Deep Studying. This new deeplearning.ai TensorFlow Specialization teaches you the best way to use TensorFlow to implement these rules as a way to begin constructing and making use of scalable fashions to real-world issues. To develop a deeper understanding of how neural networks work, we suggest that you just take the Deep Studying Specialization.

WHAT YOU WILL LEARN

• Construct pure language processing methods utilizing TensorFlow

• Course of textual content, together with tokenization and representing sentences as vectors

• Apply RNNs, GRUs, and LSTMs in TensorFlow

• Prepare LSTMs on current textual content to create unique poetry and extra

SKILLS YOU WILL GAIN

• Pure Language Processing
• Tokenization
• Machine Studying
• Tensorflow
• RNNs

Syllabus – What you’ll study from this course

Content material Score94%(12,584 rankings)

WEEK

1

7 hours to finish

Sentiment in textual content

Step one in understanding sentiment in textual content, and particularly when coaching a neural community to take action is the tokenization of that textual content. That is the method of changing the textual content into numeric values, with a quantity representing a phrase or a personality. This week you’ll study concerning the Tokenizer and pad_sequences APIs in TensorFlow and the way they can be utilized to organize and encode textual content and sentences to get them prepared for coaching neural networks!

WEEK

2

6 hours to finish

Phrase Embeddings

Final week you noticed the best way to use the Tokenizer to organize your textual content for use by a neural community by changing phrases into numeric tokens, and sequencing sentences from these tokens. This week you’ll study Embeddings, the place these tokens are mapped as vectors in a excessive dimension area. With Embeddings and labelled examples, these vectors can then be tuned in order that phrases with related that means could have the same route within the vector area. It will start the method of coaching a neural community to know sentiment in textual content — and also you’ll start by taking a look at film opinions, coaching a neural community on texts which can be labelled ‘optimistic’ or ‘unfavourable’ and figuring out which phrases in a sentence drive these meanings.

WEEK

3

7 hours to finish

Sequence fashions

Within the final couple of weeks you appeared first at Tokenizing phrases to get numeric values from them, after which utilizing Embeddings to group phrases of comparable that means relying on how they have been labelled. This gave you , however tough, sentiment evaluation — phrases akin to ‘enjoyable’ and ‘entertaining’ would possibly present up in a optimistic film evaluation, and ‘boring’ and ‘uninteresting’ would possibly present up in a unfavourable one. However sentiment can be decided by the sequence through which phrases seem. For instance, you would have ‘not enjoyable’, which after all is the alternative of ‘enjoyable’. This week you’ll begin digging into a wide range of mannequin codecs which can be utilized in coaching fashions to know context in sequence!

WEEK

4

5 hours to finish

Sequence fashions and literature

Taking all the things that you just’ve discovered in coaching a neural community primarily based on NLP, we thought it is likely to be a little bit of enjoyable to show the tables away from classification and use your information for prediction. Given a physique of phrases, you would conceivably predict the phrase most definitely to observe a given phrase or phrase, and when you’ve executed that, to do it once more, and once more. With that in thoughts, this week you’ll construct a poetry generator. It’s educated with the lyrics from conventional Irish songs, and can be utilized to provide beautiful-sounding verse of it’s personal!