What is Probabilistic Programming

What is Probabilistic Programming?

Probabilistic Programming(PP) is also a relatively unknown topic for many data scientists. However, this is an area that quickly becomes more important.

PP is a paradigm of programming that specifies probabilistic patterns and automatically deducts these patterns. It aims to bring together probabilistic modelling and traditional programming of the general-purpose to enhance and facilitate the latter. It can be used to build systems that assist decision-making in the presence of difficulty.

In this blog, I briefly discuss the specific situation of the PP. PP can be considered a tool for statistical modelling.

Randomization is the essential element of the PP and the objective of the PP is to give a statistical analysis that explains a phenomenon.

Probabilistic programming language (PPL) is built on some elements, that; we have a number of fundamental elements to generate random numbers, primitive ones to estimate probabilities & expectations and ultimately primitive ones to deduct the probabilistic inference

A PPL works a little differently than regular languages. The prior distributions are considered in the model as assumptions. The subsequent distribution of the model parameters is calculated on the basis of observed data, i.e. Inference is used to alter previous probabilities on the basis of observed data.

In PP, latent random variables enable us to model uncertainty. In this example, however, the statistical inference consists of determining the values of the latent variables.

Conceptually, Deep PPLs can express Bayesian neural networks with probabilistic weights and biases. Practically speaking, Deep PPLs have materialized as new probabilistic languages and libraries that integrate seamlessly with popular deep learning frameworks.

But now the question is this, how do you use it?

The packages like pymc3 could be the one solution by implementing the Bayesian Probabilistic Graphical models.

One way of combining deep learning with PPLs is through Deep PPLs implemented the packages of the like Tensorflow Probability (TFP). Note that, TFP is a library for probabilistic reasoning and statistical analysis. TFP is a TensorFlow-based Python package that keeps things simple to integrate probabilistic models and deep learning on modern hardware (TPU, GPU).

Note: Probabilistic programming does not involve writing software that behaves randomly.

References:

https://medium.com/swlh/a-gentle-introduction-to-probabilistic-programming-languages-bf1e19042ab6
https://www.math.ucdavis.edu/~gravner/MAT135B/materials/ch11.pdf
https://www.cs.cornell.edu/courses/cs4110/2016fa/lectures/lecture33.html
https://www.tensorflow.org/probability

Thanks for reading this blog.

Comments

SQL 5 Minutes Read

SQL CHEAT SHEET (5 Minutes Read of Structured Query Language (SQL)) SQL (Structured Query Language) is a standardised programming language used to manage relational databases and execute various operations on related data. SQL, which was developed in the 1970s, is now widely used not only by database administrators but also by developers building data integration scripts and data analysts wanting to set up and perform analytical queries. SQL is used to modify database table and index structures, add, update, and delete rows of data, and retrieve subsets of information from inside a database for transaction processing and analytics applications. Queries and other SQL operations take the form of statements, which are regularly used instructions. Select, add, insert, update, delete, create, change, and truncate are all SQL statements. In this blog, we will learn how to perform basic operations in SQL. Get function for inserting data, update data, deleting data, grouping data, or...

2. 10 FEA TURE ENCODING TECHNIQUES EVERY DATA SCIENTIST MUST KNOW FEATURE ENCODING TECHNIQUES 1- LABEL ENCODING Label encoding is intuitive and easy to understand. Label Encoding refers to converting the labels into the numeric form so as to convert them into the machine-readable form. Machine learning algorithms can then decide in a better way how those labels must be operated. It is an important pre-processing step for the structured dataset in supervised learning. Example: Suppose we have a column Height in some dataset. After applying label encoding, the Height column is converted into: where 0 is the label for tall, 1 is the label for medium and 2 is the label for short height. Limitation of label Encoding Label encoding converts the data in machine-readable form, but it assigns a unique number (starting from 0) to each class of data. This may lead to the generation of priority issues in the training of data sets. A l...

UPSC Previous year paper for AD Census Operations (Technical) & Statistical Officer (Planning/Statistics)

UPSC Previous year paper for AD Census Operations (Technical) & Statistical Officer (Planning/Statistics) Syllabus of the Test: (1) Statistical Methods (2) Sampling Techniques/Survey Methodology (3) Demography and Vital Statistics (4) Fundamentals of Applied Multivariate Analysis (5) Official Statistics (6) Basic knowledge in Computer Applications UPSC 2017 Paper For PDF file, click on the below link: https://drive.google.com/file/d/14oDfPYGI9bA2g0m-0OmfexHS8LOgaf3j/view?usp=sharing (There are 25 slides) For PDF file, click on the below link: https://drive.google.com/file/d/14oDfPYGI9bA2g0m-0OmfexHS8LOgaf3j/view?usp=sharing Any comments or suggestions are welcome. Subscribe our Youtube Channel-----> Thanks for reading this blog.

Statistics For You

Search This Blog