Rosella       Machine Intelligence & Data Mining

| Home | DataMining & Machine Learning | Products & Downloads | Site Map | Contact |

CMSR Machine Learning / Data Miner / Rule Engine Studio
Computer Vision & Data Mining & Rule Inference

CMSR Machine Learning / Data Miner / Rule Engine Studio (CMSR Studio for short) provides an integrated environment for machine learning, predictive modeling, expert system shell rule engine, and big-data data mining. CMSR is a perfect platform to develop advanced models using the deep learning technique. Developed models can be embedded on various computers and edge devices

Codingless Coding-free Machine Learning and Powerful Visualization Tools
CMSR is based on user-friendly Graphical User Interface. It's fully GUI-based. It's very easy to use and powerful. To develop computer vision models, you don't code any line of program code! In addition, you can analyze data and models very easily with powerful visualization tools.

No coding errors! No debugging! No package dependency errors!

Deploy/Embed Models on (almost) Any System
Models developed with CMSR Studio can be implemented/deployed/embedded into almost any computer system: from super computers with powerful GPUs to single board computers such Raspberry Pi, Orange Pi, Nvidia Jetson, LattePanda, etc.

Scalable High Performance Computing / Super GPU Speedup
Computer vision deep neural networks are very compute intensive. So high performance computing power is essential, especially for model training. CMSR can take advantage of big core GPU and large main-memory. CMSR can utilize large GPU cores. More GPU cores mean, more speed up. Neural networks have to be trained using randomized training data. Otherwise networks will be skewed in favor of trailing sample data. To avoid this, all training data have to be pulled into main memory. Otherwise training will take place in hugely slow pace.

CMSR is developed with Java. It runs on Windows as well as MacOS seamlessly. More importantly, it's faster than C/C++-based ML systems because of Java's Just-in-time compiler. It runs upto 2 to 3 times faster than C/C++ systems on Windows.

Computer Vision and Embedded Applications

CMSR Studio can generate computer vision model deployment program source codes for GPU in C++ and Java and Swift and Objective-C for the following GPU API platforms. Note that compute vision models are compute intensive. So GPU deployment is essential.

  • OpenCL: Java and C++. Windows, Linux, Android. SBC.
  • CUDA: Java and C++. Windows, Linux. SBC.
  • OpenGL ES (GLES3): C++. Android, Raspberry Pi. SBC.
  • Apple Metal: Swift and Objective-C. MacOS, iOS, iTablet.
  • Multiple CPU threads: Java and C++.

Generate source code. Include in your application and call. That's what all it takes! CMSR Computer Viision tasks include;

  • Image classification: Given an image, determines the type of image.
  • Image regression: Given an image, predicts a single or multiple output values.
  • Object detection: Given an image, detect objects as bounding boxes, a la YOLO.
  • Similarity regression of two images, e.g., face recognition.
  • Stereoscopic regression of two images, e.g., distance estimation.

Models incorporate pre- and post processing as follows;

  • Color inversion.
  • Color transformation and value encoding.
  • Histogram equalization.
  • Object filtering and duplicate removal.
  • Sorting.

All these are automatic upon selecting options. Generated program codes will incorporate these. So no need for extra coding. For more about computer vision on edge computing devices, see the following links;

For an example use of computer vision,
please check Skin cancer checker online app.

Predictive Modeling Algorithms and Applications

Other Algorithms and Charts

For more information, please click here.

For free downloads, please visit CMSR Download Request.


[ Screenshots of CMSR Data Miner / Machine Learning Studio ]

(For full view, click the images.)

(Example 1) RME-EP Rule Editing;
EME-EP Rule Editing

(Example 2) RME-EP Rule Analysis;
RME-EP Rule Analysis

(Example 3) CNN Image Browser;
CNN Image Browser

(Example 4) CNN Filter Visualization;
CNN Filter Visualization

(Example 5) Segmentation with Neural Clustering(SOM);
Customer Segmentation with Neural Clustering (SOM)

(Example 6) Neural Network Predictive Model;
Neural Network Predictive Modeling

(Example 7) Cramer Decision Tree Classifier;
Cramer Decision Tree Classifier

(Example 8) Variable Analysis for Modeling;
Variable Analysis for Modeling;

(Example 9) Co-items Cross-sell Analysis;
Co-items Cross-sell Analysis;

(Example 10) Data Visualization: 3D Chart;
Data Visualization: 3D Chart

(Example 11) Data Visualization: Scatterplot;
Data Visualization: Scatterplot

(Example 12) Data Visualization: Sankey Diagram;
Data Visualization: Sankey Diagram

For free downloads, please visit CMSR Download Request.

[CMSR Update] If you already have CMSR Studio, you can update to the latest version. For update download, click CMSR Update Download.


(Note that StarProbe data miner is replaced with CMSR Data Miner / Machine Learning / Rule Engine Studio.)