Data science tackles audio and insider threat analytics

US company Digital Reasoning has launched Synthesys 4, the latest version of its cognitive computing platform which leverages an ensemble of technologies including machine learning, NLP, computer vision, pattern recognition and knowledge representation.

Synthesys 4 delivers its analysis in the form of interactive data exploration tools as well as task-oriented and user-friendly Intelligent Assistants that reason over knowledge abstracted from data. The result is enhanced user productivity and continuous learning from context. 

New capabilities include: 

  • analytics enhancements leveraging deep learning neural networks across text, audio and images, as well as behavioural analytics based on anomaly detection techniques.
  • Enhanced multi-lingual capabilities, including state of the art quality for English, Spanish and Chinese with full syntactic parsing.
  • An integrated ability to conduct interactive data analytics designed for data scientists. Synthesys Notebook enables local exploration of data and import/export of results from Synthesys analytics to a range of third party databases and visualization tools.
  • Self-service model training with Synthesys Trainer provides an intuitive interface that enables non-technical users to train new analytical models according to the language and tasks that are relevant to their domain. In addition, the Annotator tool visualizes and facilitates NLP and metadata annotations in textual data.
  • A simpler yet more robust distributed architecture, making deployments quicker, more cost-effective, more extensible and easier to integrate with partner solutions. 
  • Support for Elasticsearch for high performance indexing and data exploration across multiple Hadoop distributions, including Cloudera Enterprise, HortonWorks and MapR.

According to Bill DiPietro, vice president of product management at Digital Reasoning, “Synthesys 4 will boost the adoption of cognitive computing technology by giving customers more flexibility and transparency into how the system learns from context, as well as better tools to explore the output of its cognitive algorithms and the knowledge graph.

“We are especially excited to expand into the area of Entity Behavior Analytics, combining the analysis of structured and unstructured data into a person-centric, prioritized profile that can be used to predict employees at risk for insider threats. These enriched profiles also represent valuable, holistic insights into customers, patients and physicians across use cases ranging from customer intelligence to clinical surveillance, and investigations into fraud, waste and abuse.”