Deeplearning4j: How to run AI workloads on Jakarta EE compliant servers
summary
DL4J claims an enterprise runtime that runs and monitors models on JVM servers, Android and IoT, and is already used by more than half of the Fortune 500 with largest bases in China, the US and Germany. It imports Keras/TensorFlow models (99% Keras compatibility, supports many TF ops), allocates memory off‑heap, and packs native libs (MKL/CUDA) into jars via JavaCPP/JavaCV. Example shows a Jakarta EE/MicroProfile service loading a singleton MNIST sequential model (28x28, 1 channel) and doing HTTP multipart inference.
AI-written summary · about
chapters (15)
- 0:00Intro
- 2:00About Eclipse Deeplearning4j
- 4:15Where is Deeplearning4j used
- 6:03What we see today
- 7:32Typical use cases
- 8:26The problem
- 14:03Java CPP
- 15:22MicroProfile
- 17:31Use case example
- 18:28Micro Profiles
- 20:08Singleton
- 21:30Image Conversion
- 21:50Map
- 22:32CDI
- 23:09File uploads
related talks
29mAchieve real-time inference at scale on AWS with Deep Java Library
50mApplying Machine Learning to IoT Use Cases with Deeplearning4j
44mDeep Learning in Production
50mML in Java, YES it's possible!
44mDeep Learning with DeepLearning4J and Spring Boot
46mML in Java, YES it's possible!
01h 01mDeep Learning and Java
45mJakarta EE Meets AI