DENAID: detection platform for the identification of explosives

Learning algorithms and machine learning help to increase public security.

Machine Learning for DENAID

As part of the federal government programme “Research for Civil Security”, Micromata, and other research partners investigated the possibilities of a detection platform for the detection and identification of explosives (DENAID). Research was undertaken on a technology that uses both optical methods and methods of machine learning.
We have thereby developed software that uses three different measurement methods – ion mobility spectrometry (IMS), surface-enhanced Raman spectroscopy (SERS), and surface-enhanced infra-red absorption spectroscopy (SEIRA) to identify explosives using specific algorithms. In addition to the high degree of analytical precision, it is particularly important that it is a learning system that can be trained on constantly new hazardous substances – we are referring to machine learning.

Technology and degree of innovation

The degree of innovation of DENAID is very high. The technological innovation is first in the first connection of the three measurement techniques IMS, SERS and SEIRA, which are currently considered to be the most precise. All three operate according to chemical (IMS) and spectroscopic (SERS and SEIRA) methods in order to detect and determine substances as reliably as possible. DENAID bundles the advantages and strengths of all three methods, thereby achieving increased accuracy in the identification of explosives.

Support Vector Machine

Second, the innovation potential at DENAID is based on the fact that the software is trainable. It owes its learning ability to the use of various machine learning methods. One of these is the Support Vector Machine (SVM). The mathematical method is used for IT-based pattern recognition, and enables the identification and classification of new substances. All known substances are represented as vectors in a higher dimensioned vector space and are divided by a hyperplane into classes such as “explosive” or “non-explosive”. New, unknown substances are identified by the system according to their characteristics, and are assigned to the respective class.

Target groups and marketability

The target groups of DENAID are security services, border guards and customs authorities. DENAID equips you with a technology that is equally safe, reliable, easy to use and cost effective. Compared to search dogs, for example, it offers the highest level of precision combined with low maintenance costs. It is thus the modern answer to modern challenges such as the globalisation of terrorism, alongside shrinking budgets and reduced staffing levels.

Support from the federal government

The German Federal Ministry of Education and Research (BMBF) supported the research project at runtime with EUR 1.9 million of funds from the “SME innovative” funding scheme for the fostering of cutting-edge research in small and medium-sized enterprises.

Research partners

DENAID is an interdisciplinary collaborative project. Micromata project partners are Airsense Analytics GmbH, AMO GmbH, the University of Applied Sciences and Arts (HAWK) and Laser-Laboratorium Göttingen e. V., as well as the affiliated partners Bruker Optik GmbH, the German Federal Criminal Police Office and HKS Sicherheitsservice GmbH.
The image shows on the left a PCA with linear kernel on DENAID data (SERS), and on the right a support vector machine (SVM) with radial kernel on test data.

Jule Witte

Presse & Kommunikation