R2017a at a Glance

Release Highlights

MATLAB visualizations, an app for regression, and more big data algorithms for machine learning.

Use pretrained CNN models, import Caffe models, and train with multiple GPUs and cloud instances.

New toolbox to design, simulate, and test ADAS and autonomous driving systems.

Directly run multiple parallel simulations from the parsim command.

MATLAB functions for simulating 3GPP 5G new radio technologies.

Moving to the latest release just got easier with project-wide upgrades and cross-release code integration.


Resources


Updates by Product

MATLAB Product Family

  • Interactive figure updates in the Live Editor, including title, labels, legend, and other annotations, as well as the ability to copy live script outputs to other applications​
  • heatmap chart functions for visualizing data
  • ​More functions for operating on tallarrays, including ismember, sort, conv, and moving statistics functions​​​​​

MATLAB Online

  • Use MATLAB through your web browser
  • Ideal for teaching, learning, and convenient, lightweight access
  • Available with most licenses
  • Bayesian linear regression model for analyzing the relationship between a response and a set of predictor variables
  • Vector autoregressive model for analyzing multivariate time series data including exogenous predictors
  • Web-based server management dashboard for IT configuration and control
  • Deep learning algorithms for training convolutional neural networks (CNNs) for regression tasks using multiple GPUs on PCs, on clusters, and in the cloud
  • Deep learning visualization for the features a CNN model has learned using image optimization
  • Functions for transferring weights from pretrained CNN models (AlexNet, VGG-16, and VGG-19) and models from Caffe Model Zoo
  • Regression Learner app for training regression models using supervised machine learning
  • tall array algorithms for support vector machine (SVM) and Naïve Bayes classification, bagged decision trees, and lasso regression
  • Deep learning for detecting objects using Fast R-CNN and Faster R-CNN
  • A new product for designing, simulating, and testing ADAS and autonomous driving systems

Simulink Product Family

  • parsim command for directly running multiple parallel simulations
  • Simulink Project upgrade for easily updating all files in a project to the latest release
  • Streaming for large input signals from MAT files without loading the data into memory
  • Reduced bus wiring to quickly group signals as buses and automatically create bus element ports for fewer signal lines between and within subsystems
  • Automatic port creation for adding inports and outports to blocks when routing signals
  • A new product for modeling and simulating automotive powertrain systems
  • Run-time parameters for speeding up simulation tasks and modifying component parameter values without regenerating C code
  • Onshape CAD import for using cloud-based CAD assemblies in multibody simulation

Signal Processing and Communications

  • Antenna Designer app for interactively choosing and analyzing antennas with desired characteristics​
  • Modeling and simulation for spatially-defined MIMO channels operating in multipath and scattering propagation scenarios
  • MATLAB functions for simulating 3GPP 5G new radio technologies
  • Sidelink Receive functionality for link-level simulation of LTE-A ProSe direct communications for public safety and vehicular communication applications
  • Support for generating IEEE 802.11ad compliant waveforms

Code Generation

  • Cross-release code integration for reusing model reference code generated from previous releases
  • Dynamic memory support for MATLAB Function block simulation and code generation
  • HDL code generation from IEEE standard single-precision floating-point operations
  • Support for probing and capturing internal FPGA signals to analyze in MATLAB or Simulink

Verification and Validation

  • Code checking for MISRA C:2012 Amendment 1 and new cryptography routines
  • Clone detection improvements for refactoring repeating library patterns and subsystem clones
  • DOORS® Next Generation support for linking and tracing model elements to requirements in DOORS Next Generation
  • Visualization of the effect of state activity timing on slicer highlighting for simulations
  • Support for loop and cyclic operations in MATLAB®, Simulink®, and Stateflow®

Release Notes by Product

* Indicates major update