echo "";

Artificial intelligence

Text mining, Image and Speech recognition
We actively apply AI technologies in our projects
Artificial intelligence and machine learning technologies are being increasingly used in many areas of business applications as well as consumer products.

We actively apply AI technologies in our projects, mostly focusing on image recognition and natural language processing, as well as face and speech recognition.
Вводный текст

Data processing and analysis

In our Z-monitor - a product for public procurement analysis, we used Natural Language Processing and text-mining technologies for smart search and procurement clustering.

For the initial processing of texts, we used morphological analysis, the n-gram method and statistical measures of mutual information (MI) to identify key phrases (collocations).

To obtain vector representations of collections of documents, we applied latent-semantic analysis (LSA), probabilistic latent-semantic analysis (PLSA), latent Dirichlet allocation (LDA).
Второй блок

The accuracy of the analysis

To improve the accuracy of the analysis and to reduce the effect on commonly used words, we applied the normalization of TF-IDF. Clustering of documents was carried out using the widely known k-means algorithm.
Третий блок


Recognition technology

We successfully implemented TensorFlow-based image recognition technology in Tele-Mentor product, where we solved the problem of determining and tracking the necessary equipment for medical manipulation.

We have also used VeriLook face-recognition technology (from Neurotechnology), which simplified the user authentication. In this medical exam simulator for surgeon students we have also leveraged speech recognition technologies from Google (Speech API) and Speech Technology Center.
Четвертый блок

Describing Technological Stack

We used AI technologies in both web and desktop applications.

All text-mining algorithms were implemented in python (pymorphy, numpy, and scipy libraries).

We used opencv-python and tensorflow (tensorflow-gpu) for image recognition.

For face recognition, we used VeriLook C++ SDK. 

For speech recognition and VoIP we used C++ libraries from Speech Technology Center and python-based Google Speech API.
Тех стек

AI projects

Z-monitor
A product for public procurement analysis
Medical Simulator
Tele-Mentor
Electronic equipment lab for medical students to acquire practical skills
+1 (650) 750-47-91