DevCoach 161: Machine Learning in Google Cloud | Pengenalan TensorFlow dan Ekosistemnya

Transcript

1. Machine Learning in Google Cloud Pengenalan TensorFlow dan Ekosistemnya Ridha Ginanjar Curriculum Developer
2. Machine Learning Series Series Machine Learning berfokus pada: - Mempelajari dasar Machine Learning hingga Deep Learning - Membuat dan mengoptimalkan model Machine Learning hingga Deep Learning - Mempelajari berbagai algoritma Machine Learning hingga Deep learning Machine Learning in Google Cloud Series Series Machine Learning in Google Cloud berfokus pada: - Machine Learning di lingkungan produksi - Membuat model machine learning dengan bantuan Cloud Computing - Belajar melakukan deployment model Machine Learning dengan menjadikannya sebagai API Selamat Datang di Series Machine Learning in Google Cloud
3. The Agenda • Pengenalan Machine Learning in Production • Pengenalan Machine Learning Pipeline • Pengenalan TensorFlow • Ekosistem TensorFlow
4. Pengenalan Machine Learning in Production ML in Google Cloud
5. Membangun Model Machine Learning Adalah Sebuah Inti Dari Produk ML Data Collection Data Preparation Choose and train the model Model Evaluation
6. Tapi, membangun model ML hanya 5% dari seluruh Machine Learning system ML Code Analysis Tools Process Management Tools Feature Engineering Data Collection Data Extraction Machine Resource Management Configuration Serving Infrastructure
7. Machine Learning Pipeline ML in Google Cloud
8. Staging/pre-production/production environment Raw Data Data Extraction Data Analysis Data Preparation Model Training Model Evaluation Model Validation Trained Model Model Registry Prediction Service Performance Monitoring Experimentation/development/test environments
9. Staging/pre-production/production environment Raw Data Data Extraction Data Analysis Data Preparation Model Training Model Evaluation Model Validation Trained Model Model Registry Prediction Service Performance Monitoring Experimentation/development/test environments
10. Staging/pre-production/production environment Raw Data Data Extraction Data Analysis Data Preparation Model Training Model Evaluation Model Validation Trained Model Model Registry Prediction Service Performance Monitoring Experimentation/development/test environments
11. Staging/pre-production/production environment Raw Data Data Extraction Data Analysis Data Preparation Model Training Model Evaluation Model Validation Trained Model Model Registry Prediction Service Performance Monitoring Experimentation/development/test environments
12. Staging/pre-production/production environment Raw Data Data Extraction Data Analysis Data Preparation Model Training Model Evaluation Model Validation Trained Model Model Registry Prediction Service Performance Monitoring Experimentation/development/test environments
13. Staging/pre-production/production environment Raw Data Data Extraction Data Analysis Data Preparation Model Training Model Evaluation Model Validation Trained Model Model Registry Prediction Service Performance Monitoring Experimentation/development/test environments
14. Staging/pre-production/production environment Raw Data Data Extraction Data Analysis Data Preparation Model Training Model Evaluation Model Validation Trained Model Model Registry Prediction Service Performance Monitoring Experimentation/development/test environments
15. Staging/pre-production/production environment Raw Data Data Extraction Data Analysis Data Preparation Model Training Model Evaluation Model Validation Trained Model Model Registry Prediction Service Performance Monitoring Experimentation/development/test environments
16. Staging/pre-production/production environment Raw Data Data Extraction Data Analysis Data Preparation Model Training Model Evaluation Model Validation Trained Model Model Registry Prediction Service Performance Monitoring Experimentation/development/test environments
17. Staging/pre-production/production environment Raw Data Data Extraction Data Analysis Data Preparation Model Training Model Evaluation Model Validation Trained Model Model Registry Prediction Service Performance Monitoring Experimentation/development/test environments
18. Staging/pre-production/production environment Raw Data Data Extraction Data Analysis Data Preparation Model Training Model Evaluation Model Validation Trained Model Model Registry Prediction Service Performance Monitoring Experimentation/development/test environments
19. Staging/pre-production/production environment Raw Data Data Extraction Data Analysis Data Preparation Model Training Model Evaluation Model Validation Trained Model Model Registry Prediction Service Performance Monitoring Experimentation/development/test environments
20. TensorFlow ML in Google Cloud
21. Apa itu TensorFlow?
22. Lalu, apa itu Tensor?
23. Bagaimana Cara Kerjanya?
24. Yang ada di dalam Suatu Function
25. Di balik ML Model Traditional Programming Machine Learning Rules Data Answers Rules Data Answers
26. Di balik ML Model Traditional Programming Y= 2x + 1 X = -1 Y = -1 Data yang diketahui Rules Answers , 0, 2 , 3, 4 , 1 , 5 , 7, 9
27. Di balik ML Model Traditional Programming Machine Learning Y= 2x + 1 X = -1 Data yang diketahui Y= 2x + 1 X = -1, 0, 2, 3, 4 Data yang diketahui Y = -1, 1, 5, 7, 9 Rules Rules Answers Answers , 0, 2 , 3, 4 Y = -1 , 1 , 5 , 7, 9
28. Bagaimana cara Menggunakan TensorFlow?
29. Ekosistem TensorFlow ML in Google Cloud
30. Apa itu Ekosistem TensorFlow? TensorFlow (Core) TensorFlow.js (For Web) TensorFlow Lite (For Mobile & Edge) TensorFlow Extended (TFX) (For Production) Membuat machine learning model menggunakan Python. Membuat dan menjalankan ML model di lingkungan web dengan JavaScript. Menjalankan inference di lingkungan mobile dan embedded device seperti Android, iOS, hingga Raspberry Pi. Menjalankan Machine Learning Pipeline di lingkungan produksi seperti Google Cloud.
31. Model yang dihasilkan Setiap Ekosistem TensorFlow (Core) TensorFlow.js (For Web) TensorFlow Lite (For Mobile & Edge) Format Model: - .keras format - SavedModel - HDF5 Format Model: - TensorFlow.js Model Format Format Model: - FlatBuffer
32. The Inference ML in Google Cloud
33. Bentar, apa itu Inference? Inference adalah fase prediksi dimana suatu model yang berhasil dilatih digunakan untuk memprediksi data baru yang belum pernah dilihat sebelumnya (selama fase training) Model Data Baru (Input) Hasil Prediksi (Output)
34. Cara melakukan inferensi Menggunakan Lokal Komputer
35. Terima Kasih. Eitss … Kita punya informasi khusus untuk series ML in Google Cloud selanjutnya
36. Staging/pre-production/production environment Raw Data Data Extraction Data Analysis Data Preparation Model Training Model Evaluation Model Validation Trained Model Model Registry Prediction Service Performance Monitoring Experimentation/development/test environments
37. Kita akan fokus terhadap 2 hal berikut Trained Model Prediction Service
38. Feedback! Hadiah: • 1 Token Langganan Academy (30 Hari) *untuk pengisi feedback terpilih! dicoding.id/devcoachfeedback