All Tools
D
OtherFreeOpen Source
DEEPLEARNING4J
Deep learning for Java, Scala, and the JVM ecosystem
Apache-2.0
ABOUT
Most deep learning frameworks are Python-first, leaving Java and JVM ecosystems — where most enterprise production systems run — with few options for native model training and inference. DeepLearning4J (DL4J) brings deep learning to the JVM with native Java APIs, model import from Keras, TensorFlow, and ONNX/PyTorch via a modular C++ math library (libnd4j), and SameDiff — a PyTorch-like autodiff library. It integrates with Hadoop, Spark, and Kafka for distributed training on big data pipelines, and runs on CPUs and GPUs without requiring Python or native library conflicts in JVM deployments.
INTEGRATION GUIDE
1. Train an LSTM for time-series forecasting on financial data directly in a Java trading application
2. Import a Keras image classifier and serve it through a Spring Boot REST API
3. Run distributed deep learning training across a Spark cluster on Hadoop data
4. Build a real-time anomaly detection pipeline in Apache Kafka Streams using DL4J
5. Deploy a TensorFlow model in a JVM microservice without any Python dependencies
TAGS
javadeep-learningjvmscalaneural-networktensorflowonnxpytorch