Research powered by Deep Netts AI

Explore how research teams use Deep Netts to bring deep learning into the Java software they already work with.

Nuclear Physics

Predicting particle trajectories with neural networks

Tracking took 80–90% of event processing time in the CLAS12 detector, and at high beam intensity good tracks got lost among thousands of candidate combinations.

+15.6%
reconstructed tracks on real CLAS12 data
35%
faster track reconstruction
+15–35%
events in two- and three-particle reactions
Read case study
Drug Discovery

Identifying promising compounds for drug discovery

Screening produces tens of thousands of data points per session – but deep learning tools built on native libraries are hard to run on specific hardware, and QSAR methods need structural data many compounds lack.

Pure JVM
no native libraries to install
No QSAR input
models learn from properties already in the compound library
Read case study

Building something similar in Java?