Programming language Scala combines object-oriented and functional programming. It is an extension of Java and runs on Java Virtual Machine (JVM). Many developers prefer Scala over Java since the same programmes can be written on the former using a significantly smaller number of lines.
Scala’s complex features aids better coding and offers efficient performance. Scala integrates functional programming and object-oriented programming into one.
REGISTER FOR OUR UPCOMING ML WORKSHOP
We have listed eight Scala libraries for data scientists to use:
Built on Spark, Apache Spark is a scalable machine learning library. It consists of algorithms and utilities including the likes of regression, classification, collaborative filtering, clustering, dimensionality reduction and underlying optimisation primitives.
The MLlib library provides functional APIs for Java, Python and R, and consists of two different packages:
- MLlib: The RDD-based library comprising machine learning algorithms and utilities including the likes of regression, classification, collaborative filtering, clustering, dimensionality reduction and underlying optimisation primitives. It can be used for basic statistics including correlations, hypothesis testing and random data generation. Additionally, its utilities include linear algebra and data handling.
MLlib fits into Spark’s APIs and interoperates with NumPy in Python and R libraries. MLlib’s goal is to make practical machine learning scalable and easy.
- ML: The more recent package introduced in Spark 1.2, provides high-level APIs to help users create practical machine learning pipelines. It operates on data frames and datasets.
DeepLearning4J or DL4J is an open-source, distributed, deep learning library for Java and Scala. Created by the Eclipse team, DL4J takes advantage of the latest distributed computing frameworks– Apache Spark and Hadoop to accelerate training.
The libraries are open-source and maintained by the Konduit team. While DL4J is written in Java, it is compatible with JVM languages– Scala, Clojure and Kotlin. Its underlying computations are written in C, C++ and Cuda.
DL4J allows developers to create deep neural nets (called ‘layer’) from various shallow nets, allowing them to combine variational autoencoders, sequence-to-sequence autoencoders, convolutional nets or recurrents nets, as required.
Intel tool BigDL is a distributed deep learning library for Apache Spark that can be used to write deep learning applications as standard Spark programmes, directly running on top of Spark or Hadoop clusters. It provides support for deep learning including numeric computing and high-level neural networks. Additionally, developers can also load pre-trained Torch or Caffee models into Spark programmes with the use of BigDL.
It can be scaled to perform data analytics by leveraging Apache Spark. Additionally, BigDL allows the implementation of synchronous SGD and all-reduce communications on Spark.
An integration of H2O’s fast, scalable machine learning engine with Spark by Sparkling Water provides utilities to publish Spark data structures (RDDs, DataFrames and Datasets) as H2O’s frames, and vice-versa. Additionally, H2O Sparkling Water provides DSL to use Spark data structures as input for H2O’s algorithms; basic building blocks to create machine learning applications; and Python interface to enable use of Sparkling Water directly from PySpark.
Etsy-created Conjecture is a framework to build machine learning models in Hadoop using Scalding DSL. It helps in the development of statistical models as viable components. Conjecture’s applications include classification and categorisation, recommender systems, ranking, filtering and regression.
Conjecture focuses on flexibility and can handle a wide range of inputs. Its integration with Hadoop and Scalding DSL enables the seamless handling of humongous amounts of data.
Written in Scala, Akka toolkit is used to build concurrent, distributed and resilient message-driven applications for both Scala and Java.
Akka is often called an actor-based model where the ‘actor’ actually is similar to an object in an object-oriented model. The only difference is that unlike object-oriented models, an actor-based model is specifically designed and architected to serve as a concurrenct model. It created a layer between the actors and the underlying system.
Recently, Akka launched Akka 2.6.15 with 21 closed issues. It adheres to the Reactive Manifesto. It is event-driven, scalable, resilient and responsive.
As a suite of lightweight library by Scana Spray provides client and server-side REST and HTTP support on top of Akka.
Ever since its inception, rather than building application cores, Scales has been focused on providing tools for building integration layers. It guarantees asynchronous, non-blocking, actor-based and high performance request processing. The internal Scala DSL provides a defining web service behaviour, efficient and convenient testing capabilities.
Sacala Language-Integrated Connection Kit or Slick is a modern database query and access library for Scala. It allows developers to work with stored data, almost as if using Scala collections. This, while giving full control to the developer over when a database access happens and which data is transferred.
Slick also features an extensible query compiler to generate code for different backends. It creates and executes database queries for H2, MySQL, and PostgreSQL.
Join Our Telegram Group. Be part of an engaging online community. Join Here.
Subscribe to our Newsletter
Get the latest updates and relevant offers by sharing your email.
Credit: Google News