Our system is the partner that expert Compliance officers dream about. Detectica’s artificial intelligence system learns from analyst activities. Built and tuned for compliance issues ranging from financial regulatory violations to HR violations, our system utilizes human-machine loops, behavioral scoring, deep learning, and a combination of other proven techniques from the field of data science. Our system is built to deter bad behavior, while detecting egregious compliance violations.

The Detectica system:

  • Can quickly connect to and process data from major industry-standard systems such as Exchange, Bloomberg, and others.
  • Already incorporates models to detect ethical and financial regulatory violations.
  • Analyzes English language communications including transcriptions of voice.
 

DETECTICA SUPPORTS COMPLEX INVESTIGATION WITH SEAMLESS AI

Detectica-system.jpg

We specialize in incorporating expert knowledge – in the form of analyst interactions with Detectica - to continuously improve models. As AI continues to mature, we are seeing that such systems can only be accurate to a maximum of 80%.  Human-in-the-loop learning uses experts to continue tuning to that top 20%. These loops also ensure that we can capture unexpected changes in behavior and outliers that only an expert compliance officer would understand

We specialize in incorporating expert knowledge – in the form of analyst interactions with Detectica - to continuously improve models. As AI continues to mature, we are seeing that such systems can only be accurate to a maximum of 80%.  Human-in-the-loop learning uses experts to continue tuning to that top 20%. These loops also ensure that we can capture unexpected changes in behavior and outliers that only an expert compliance officer would understand

We have deep and broad experience building systems for fraud and behavioral detection. We are widely published, read, and understand the nuances of applying machine learning and AI to compliance.


Publications


Patents

Task-Agnostic Integration of Human and Machine Intelligence

JM Attenberg, PG Ipeirotis

US Patent 20,170,053,215

 

Methods, systems, and media for identifying errors in predictive models using annotators

JM Attenberg, PG Ipeirotis, FJ Provost

US Patent 9,311,599