Signal processing can be a hefty job as it cuts across several processes in order to get the job done. Variables like statistical analysis, data mining, virtualization, design, preprocessing, training of neural networks and adaptive systems, and the deployment of them, amongst many others come into play. As such, this calls for heavily sophisticated tools in order to get the job done. This calls for applications like Peltarion Synapse, an aptly built, sophisticated tool that is very handy for this purpose.
Peltarion Synapse is a component-based development environment for neural networks and adaptive systems. The first version of this product was released in May 2006, created by Peltarion. The application allows for data mining, statistical analysis, visualization, preprocessing, design and training of neural networks and adaptive systems and the deployment of them. It utilizes a plug-in based architecture making it a general platform for signal processing.
How Can Peltarion Synapse Make Life Easier for You?
Because of its plugin-based design, this allows Synapse to be very general. This application software is based on Microsoft .NET’s framework. Also, all of Synapse components are .NET components as well. However, even though Peltarion is yet to release an official API for the Synapse platform, concerned users have gone ahead to make custom user-made components. The only downside here is that while some of them strive to keep it original, demonstrating the openness of the platform, others simply do it for the sake of it and hardly pay attention to such details.
Unlike some, the development cycle in Synapse is adapted from the canonical data mining cycle. However, a noteworthy difference with this software is that in Synapse, the cycle is not in any way linear. Even so, it supports the iterative approach where users can freely move between the steps and get things done faster. Basically, Synapse features four different operating modes which make up its entire development cycle.
The preprocessing mode is handled by data mining and data preparation. In this mode, the users are allowed to import, visualize, explore, and transform data using an array of options. – Data can be imported through the use of format components (the standard version includes preset tools for reading and writing data from CSV (text) files, SQL databases, images, and XML). Imported data can be visualized through visualizer components. Also, filters can be applied to the data according to the user's specifications (the filter components range from simple data rearrangement to more advanced FFT and outlier removal filters). Another notable feature is the fact that Synapse’s visualizers include a wide range of plots and grids; these grids/plots can be interconnected and pronged out to perform complex data mining tasks.
In the design, mode components are linked so that they construct a topology. When this is done successfully, the linked components make it possible to have a signal flow. By so doing, a pipe filter machine is created; when a signal is set on a component, it filters the signal. Afterward, the filtered signal can then be piped to the next element on the linked chain of components which make up the topology (components can be either static or adaptive). Apart from the regular filters, there could also be sources or sinks (like plots or data loggers. There is a variety of components that are accompanied with the standard version of Synapse, they include simple neural network components such as weight layers and function layers, up to whole neural networks such as self-organizing maps and even a lot more complex array of static elements including the fuzzy logic component. Also, the control system is usually selected and defined in the design mode.
The training mode is used for adapting your system, (or in a more general scope, for starting the control system that regulates the information flow). Visually, this mode is quite similar to the design mode, especially because it displays the same components. Additionally, the training mode allows for the execution of high-level optimizers, including genetic algorithms, particle swarm optimization, and simulated annealing. Remote execution and training are also possible in this mode.
The postprocessing mode is designed for analyzing a trained system and also for the preparation of such a system for optimal performance on end-user. Here, system performance can be tested with the use of statistical analysis. Sensitivity analysis (the sensitivity of the input-output relations of a system) can also be carried out, and reports can be generated as well.
The deployment component being the final component, allows for the export of a system made in Synapse to a single .NET component. Here, the system in the development environment will be downgraded so that it only contains the minimal system requirements needed for successful execution. Afterward, it is then compiled into an assembly. (this assembly can then be used in any .NET framework or .NET Compact Framework application). This also allows for the deployment to other embedded devices.
Key Features of Peltarion Synapse Include:
- Comprehensive user interface.
- Modern tools.
- Full support through preprocessing to postprocessing and deployment.
- Four distinct modes.
- Regular update.
- Application is user-friendly.
- Ease of use.
This application is no doubt sophisticated and capable of seeing things through from scratch to deployment. It offers some of the latest modern tools to work with. And though the file size and system requirements are on the high side, it makes up for this with top class functionality.
We don't have any change log information for Synapse 1.2 yet. If you have any change log info for this version of Synapse you can share with us.
Last Updated: 2019-08-29
File size: 9.34 MB
Operating system: Windows 10, Windows 8/8.1, Windows 7, Windows Vista, Windows XP
MD5 Checksum: a36dd2582bf2d3f4c1e5b9a992cb326d