Wolfram Mathematica is a calculation software published by Wolfram Research since 1988 and used in scientific circles to perform algebraic calculations and create programs. Wolfram Research has released version 12.1 of its programming language Wolfram Language as well as Wolfram Mathematica.

The amount of new features detailed in a blog post by creator Stephen Wolfram is quite extensive and has a lot to offer those who use any of the products to get started with machine learning. Version 12.1, for example, saw the addition of Julia and R to the collection of external languages, which means that system capabilities should now be more accessible to data scientists.

GAN, BERT, GPT-2, ONNX: the latest advances in machine learning

Users who regularly use the Wolfram neural network repository will find 25 new types of networks, including the popular BERT language representation model and the Generative Pretrained Transformer 2 which is used for text generation systems. The system now also comes with a symbolic NetGANOperator and a TrainingUpdateSchedule option, which are intended, for example, to allow general NetTrain functions in Wolfram to work with generative conflicting networks like those often used in unsupervised learning or by enhancement.

Other than that, importing implementations of new neural networks should become a little easier in the future, as version 12.1 now supports ONNX, an open format for representing machine learning models. Those working in image processing get more help with additions like FindImageText, which detects text in an image and marks it, while audiophiles will take advantage of SpeechInterpreter and SpeechCases.

The start of video processing

If this dimension is not already sufficient, the new versions are the first to actively support calculations on video material. Developers of complex recognition systems can finally use machine learning or image processing capabilities on video, or do simpler things like extracting sequences of images or calculating all kinds of average values.

Other highlights

The Wolfram team also reworked data sets. Users will be pleased to learn that they can now set default values ​​for the number of rows and columns to display and that they can now better control the appearance of a dataset. Data beyond what can be seen in a notebook is stored directly in the notebook in the new version, which means that it will always be accessible when reopened. Meanwhile, two-dimensional data can now be captured and viewed using the TableView experimental function.

Source : Wolfram