Commenting David Rich, MATLAB marketing director said, “One of the key challenges in moving AI from hype to production is that organisations are hiring AI ‘experts’ and trying to teach them engineering domain expertise. With R2019a, MathWorks enables engineers to quickly and effectively extend their AI skills, whether it’s to develop controllers and decision-making systems using reinforcement learning, training deep learning models on NVIDIA DGX and cloud platforms, or applying deep learning to 3-D data,”
R2019a introduces the Reinforcement Learning Toolbox that further enhances the MATLAB workflow for AI. The toolbox looks to help develop a type of machine learning that trains an “agent” through repeated trial-and-error interactions with an environment to solve controls and decision-making problems.
The toolbox reiterates the company’s commitment to AI, building on last year's R2018b introduction of the Deep Learning Toolbox, which was enhanced with support for NVIDIA GPU Cloud, Amazon Web Services, and Microsoft Azure, and interoperability through support of the ONNX exchange format.
Further support for AI in R2019a includes significant enhancements to the Computer Vision Toolbox, Data Acquisition Toolbox and the Image Acquisition Toolbox.
The R2019a also features several new signal processing and communications products to support wireless and electronics development, including:
The Mixed Signal Blockset – a Simulink add-on providing fast model construction, rapid simulation, and deep insights into mixed-signal system design models with dedicated analysis and visualization tools
SerDes Toolbox – a Simulink add-on that offers the SerDes Designer app for rapid design, analysis, and modeling of wired communications transmitters and receivers
SoC Blockset – a Simulink add-on that enables simulation and exploration of FPGA, ASIC and SoC architectures, cosimulation of algorithms and hardware platforms, and performance monitoring and bottleneck detection
R2019a also makes a number of notable advances in the company’s Polyspace static analysis product family. This includes new products that support enterprise-scale use of these products for the design and development of safety- and business-critical software.
