Download to read offline
Uploaded byDeNA
課題ドリブン、フルスタックAI開発術 [MOBILITY:dev]
The document outlines an agenda for a presentation on full stack AI development by Kosuke Kuzuoka, who discusses his background and experience in AI research. It provides an overview of object detection in computer vision, the AI development pipeline, and the advantages of using serverless architecture for automation and improved efficiency. The presentation introduces an internal tool called Kaiseki-kun that automates various processes, enabling developers to focus on AI work and model training.
More Related Content
Similar to 課題ドリブン、フルスタックAI開発術 [MOBILITY:dev]
![[DSC Europe 24] Thomas Kitzler - Building the Future – Unpacking the Essentia...](https://cdn.slidesharecdn.com/ss_thumbnails/thomaskitzler-241220214738-670777be-thumbnail.jpg?width=600ounds&width=560&fit=bounds)
![タクシーxAIを支えるKubernetesとAIデータパイプラインの信頼性の取り組みについて [SRE NEXT 2020]](https://cdn.slidesharecdn.com/ss_thumbnails/202001srenext-200125032719-thumbnail.jpg?width=600ounds&width=560&fit=bounds)
![後部座席タブレットにおけるMaaS時代を見据えた半歩先のUX設計」 [MOBILITY:dev]](https://cdn.slidesharecdn.com/ss_thumbnails/20191031mobilitydevyoneyama-191031085554-thumbnail.jpg?width=600ounds&width=560&fit=bounds)
![ドライブレコーダ映像からの3次元空間認識 [MOBILITY:dev]](https://cdn.slidesharecdn.com/ss_thumbnails/20191031mobilitydevmiyazawa-191031085336-thumbnail.jpg?width=600ounds&width=560&fit=bounds)
![MOVで実践したサーバーAPI実装の超最適化について [MOBILITY:dev]](https://cdn.slidesharecdn.com/ss_thumbnails/20191031mobilitydevera-191031084650-thumbnail.jpg?width=600ounds&width=560&fit=bounds)
![DeNAのQCTマネジメント IaaS利用のベストプラクティス [AWS Summit Tokyo 2019]](https://cdn.slidesharecdn.com/ss_thumbnails/aws-summit-tokyo-2019-kei-190617011042-thumbnail.jpg?width=600ounds&width=560&fit=bounds)
課題ドリブン、フルスタックAI開発術 [MOBILITY:dev]
- 1. Full Stack AIDevelopment Kosuke Kuzuoka AI System Group DeNA Co., Ltd.
- 2. Agenda 2 Introduction AI Development Pipeline GoingServerless! 1 3 Wrap Up4 2
- 3.
- 4. About me • Profile •Kosuke Kuzuoka • 23 years old • Experience • June 2018 – Present AI Research Engineer DeNA Co., Ltd. • March 2017 – June 2018 R&D Manager Photoruction, inc. • Interests • Self Driving Cars • Computer Vision • Cars, especially Tesla 4
- 5. 5 Brief Intro toObject Detection • An active research area among computer vision community • Task is detecting objects (like cats) in an image • Modern algorithms heavily rely on deep learning • Modern algorithms have million of parameters Photo by Paul Hanaoka on Unsplash
- 6. Brief Intro toObject Detection 6 • Detector takes an image, find object-ish regions and determines what the object is • Detector has hyper-parameters, that need to be carefully tuned • Detector learns through propagating losses backwards • Learning takes hours on modern GPUs Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks https://arxiv.org/abs/1506.01497
- 7. 7 A cat isdetected as a cat, hence it’s a true positive. Wrongly detected as cats, hence they are false positives Photo by Paul Hanaoka on Unsplash
- 8.
- 9. AI Development Pipeline 9 1.Train, validate and test AI model on GPU machine 2. Check the mAP (evaluation protocol), visualize results etc.. 3. Adjust hyper-param then go back to 1.
- 10. AI Development Pipeline 10 1.Train, validate and test AI model on GPU machine 2. Check the mAP (evaluation protocol), visualize results etc.. 3. Adjust hyper-param then go back to 1.
- 11. Problems 11 • Error-prone process(misspelling commands, etc.) • Going back and forth between EC2 instances… • Inefficient process, like drawing boxes, uploading to third party app for visualization etc.
- 12. Solutions 12 • Work harderand harder... • Automation via workflow engine • Internal tool to evaluate, visualize and analyze models
- 13. Solutions 13 • Work harderand harder... • Automation via workflow engine • Internal tool to evaluate, visualize and analyze models
- 14.
- 15. Why Serverless? 15 • Easyto deploy and maintain • You can think about servers, less ☺ • New features can be easily added • Collaborations made easy • You can use python Image source: https://serverless.com/
- 16. Introducing Kaiseki-kun 16 • Maintainableand scalable analysis tool • Automates boring manual processes • Visualization through a user-friendly web viewer • Lets you focus on AI work (parameter tuning, etc.)
- 17. Kaiseki-kun Architecture 17 • Predictionsfrom GPU instance • Evaluates predictions and stores them in DynamoDB • Secured API through a user friendly web viewer
- 18. Kaiseki-kun Architecture 18 • Predictionsfrom GPU instance • Evaluates predictions and stores them in DynamoDB • Secured API through a user friendly web viewer
- 19. Kaiseki-kun Architecture 19 • Predictionsfrom GPU instance • Evaluates predictions and stores them in DynamoDB • Secured API through a user friendly web viewer
- 20. 20 5 true positivesand 1 false negative. Adjusting the threshold further...
- 21. 21 The model wasn’tconfident enough. Perhaps add more data on the category?
- 22. Serverless is Awesome! 22 •Developed in only two weeks! • Adding more features using spare time (while training models, etc.) • Task driven development enables work on both AI and full-stack
- 23.
- 24. Wrap up 24 • Serverlessenables us develop both AI model and full-stack • Boring process like visualizing outputs and evaluating models are automated • More features on their way... • Work smarter, not harder ☺
- 25. Becoming a Full-StackAI Engineer like.. Photo by Tran Mau Tri Tam on Unsplash Photo by Stephanie LeBlanc on Unsplash
- 26. 26 Thanks! Reach me at: Facebook,LinkedIn, Medium