Python is now one of the most popular and widely used programming languages, and has been overtaken by many programming languages in the industry. It is popular among developers for a number of reasons, not the least of which is the large number of libraries available to users. Python’s ease of use and flexibility have attracted many developers to create new libraries for machine learning. There is a library that everyone will be talking about, and that’s TensorFlow, but I won’t talk about it here. So, the following is today’s dry goods, you can also share their favorite library in the comments ~

1.Keras

Keras is a machine learning API written in Python that runs on TensorFlow, the top platform for machine learning. The advantage is that the network model can be realized quickly, data input and output is also very convenient, so that you can focus on the network model itself, suitable for novices. The biggest disadvantage is slow! As a high-level API, reasoning speed and so on certainly not tf, MXNET those faster.

2. PyTorch

Features: Handles n-dimensional tensors, similar to Numpy, but runs on gpus. Supports automatic differentiation to build and train large neural networks.

3. fastai

By taking advantage of current best practices, FASTAi simplifies the training process and speeds up neural networks very quickly. A single API covers almost all common deep learning applications.

4. JAX

Jax is a combination of Autograd and XLA to provide high-performance machine learning research. As an updated version of Autograd, JAX can automatically differentiate native Python and Numpy functions. Differentiates with closures of loops, branches, recursion and closures, and differentiates derivatives of derivatives. Reverse mode differential through grad is supported.

5. FastText

It is a library that allows you to effectively learn word meanings and sentence classification.

6. spaCy

SpaCy V3.0 has all the new Transformer based pipes to bring spaCy accuracy up to SOTA. You can use any pre-trained Transformer to train your own pipes, or you can share Transformers between multi-components and multi-tasks. SpaCy’s Transformer supports integration with the PyTorch and HuggingFace Transformers libraries and has access to many pre-trained models in the pipeline.

7. gensim

It uses a large corpus for subject modeling, document indexing, and similarity retrieval. The target audience is the natural language processing (NLP) and information retrieval (IR) communities.

8. NLTK

It is a natural language toolkit, a set of open source Python modules, data sets, and tutorials for the research and development of natural language processing.

9. TextBlob

Simple, Python-style, is a library for processing textual data. It provides a simple API for diving into common natural language processing (NLP) tasks such as pos tagging, noun phrase extraction, sentiment analysis, classification, translation, etc.

10. Pillow

It is a very user friendly branch of PIL. PIL is the Python image library.

11. OpenCV

Open source computer vision library.

12.LightGBM

It helps developers build new algorithms using a redefined base model known as a decision tree. LightGBM features: fast calculation speed, high production efficiency, intuitive and easy to use. This library provides a highly extensible, optimized, and fast implementation of gradient enhancement, which makes it popular among machine learning developers.

13.Pandas

Pandas is a machine learning library in Python that provides advanced data structures anda wide variety of analysis tools. Primarily, the ability to transform complex data operations using one or two commands. It also has many built-in functions for grouping, data composition, filtering, and time series functions. When used with other libraries, Pandas ensures high performance and good flexibility.

14.. Numpy

The array interface is Numpy’s best feature. It is very easy to understand and use, making complex mathematical implementations very simple. It is widely used and therefore has many open source contributors. This interface can be used to represent images, sounds, and other binary raw streams as n-dimensional real arrays.

15.Click

Click was developed to create a beautiful command line interface in a composable way with minimal code. It is intended to make the process of writing a command line tool quick and fun, while preventing any problems that might result from not implementing the desired CLI API. There are three main features: arbitrary nesting of commands, automatic help page generation, and support for lazy loading of subcommands at run time.

The following xiaobian built **Python Learning Exchange Group**, welcome to exchange and study.