Machine Learning Research in blibli

Presentation slide for bukatalks feat JVM Meet Up at Bukalapak

Transcript

1. Disclaimer Presentations are intended for educational purposes only and do not replace independent professional judgment. Statements of fact and opinions expressed are those of the participants individually and don’t necessarily reflect those of blibli.com. Blibli.com does not endorse or approve, and assumes no responsibility for, the content, accuracy or completeness of the information presented.
2. Machine Learning Research in blibli Hendri Karisma hendri.karisma@gdn-commerce.com / hendri.k.id@ieee.org
3. Hendri Karisma • Sr. Research and Development Engineer at blibli.com (PT. Global Digital Niaga) • Rnd Team for Data Science/intelligence system • Working for Fraud Detection System. Current working in dynamic recommendation system project.
4. Definition of Informatics “Automation of Information” – Prof. Dr. Ing. Iping Supriana
5. Machine Learning Definition “A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P, if its performance at tasks in T, as measured by P, improves with experience E.” – Prof. Tom Mitchel
6. How it works
7. Machine Learning #1 • Supervised • Unsupervised • Reinforcement Learning • Semi-Supervised • Deep Learning
8. Machine Learning #2
9. Machine Learning Taxonomy
10. JVM Tech (tools) for ML • weka • Deeplearning4j (working with spark and GPU) • H2O (working with spark and GPU, support Tensorflow, MxNet, and cafe) • jcuDNN (JNI for wrapping nVidia cuDNN) • Mahout • MLlib spark
11. Artificial Intelligence in Industry • Fraud Detection System • Dynamic Recommendation System and User Profiling • Traveling Salesman Problem and Binpacking Problem for better warehouse management • Social Media Analysis • Chatbot • Company condition forecasting • Governance simulation
12. The Complexity #1
13. The Complexity #2 • Big data : volume, variety, velocity, and veracity. (You might consider a fifth V, value.) • Knowledge representation or the architecture of the model • Unimplemented methods/algorithms in any libraries • Stack of methods • Data mostly unlabeled data • Data resources (microservices) • Features Engineering (especially from unstructured data) • Machines (Hardware) • High Performance Computing
14. Stack of Methods • More complex methods and models • Methods characteristic & behavior • Methods customization • Ex. Semi-supervised, Deep learning, features engineering • Sample cases : our research in FDS and dynamic recommendation system
15. High Performance Computing https://hadoop4usa.wordpress.com/2012/04/13/scale-out-up/
16. Machines • Cluster machine • GPU machines (OpenCL and nvidia CUDA)
17. High Performance Computing #2 • In-memory data fabric: provides low-latency access and processing of large quantities of data by distributing data across the dynamic random access memory (DRAM), Flash, or SSD of a distributed computer system.
18. Machine Learning Flow Scala + Play + Akka + Mongo The problem is microservices
19. Next Machine Learning Flow Scala + Play + Akka + Mongo + kafka
20. New Engine Machine Learning • Kafka • Tensorflow (utilize GPU) • Spark • Java/Scala/python
21. Dynamic Collaborative Filtering using HMM
22. Topological Data Analysis • Coordinate invariance • Deformation Invariance • Compressed Representations
23. Averi Project AVERI PROJECT twitter porn account detection
24. Example AVERI PROJECT twitter prostitution account detection
25. THANK YOU Any question?