Presentation slide for AI seminar at Artificial Intelligence Research Center, The National Institute of Advanced Industrial Science and Technology, Japan.
URL (in Japanese): https://www.airc.aist.go.jp/seminar_detail/seminar_046.html
This document discusses generative adversarial networks (GANs) and their relationship to reinforcement learning. It begins with an introduction to GANs, explaining how they can generate images without explicitly defining a probability distribution by using an adversarial training process. The second half discusses how GANs are related to actor-critic models and inverse reinforcement learning in reinforcement learning. It explains how GANs can be viewed as training a generator to fool a discriminator, similar to how policies are trained in reinforcement learning.
SAM is a new segmentation model that can segment objects in images using natural language prompts. It was trained on over 1,100 datasets totaling over 10,000 images using a model-in-the-loop approach. SAM uses a transformer-based architecture with encoders for images, text, bounding boxes and masks. It achieves state-of-the-art zero-shot segmentation performance without any fine-tuning on target datasets.
This document discusses generative adversarial networks (GANs) and their relationship to reinforcement learning. It begins with an introduction to GANs, explaining how they can generate images without explicitly defining a probability distribution by using an adversarial training process. The second half discusses how GANs are related to actor-critic models and inverse reinforcement learning in reinforcement learning. It explains how GANs can be viewed as training a generator to fool a discriminator, similar to how policies are trained in reinforcement learning.
SAM is a new segmentation model that can segment objects in images using natural language prompts. It was trained on over 1,100 datasets totaling over 10,000 images using a model-in-the-loop approach. SAM uses a transformer-based architecture with encoders for images, text, bounding boxes and masks. It achieves state-of-the-art zero-shot segmentation performance without any fine-tuning on target datasets.
ソフトウェア業界ではワクワクする新しいテクノロジーがどんどん生まれ、それが世の中で使われるまでも早くなっています。2018年に革新があった Deep Learning は、既に民主化・日常化もしてます。この講演では、そのソフトウェアの今を俯瞰し、今後どうなっていくのか? その未来予想とともに。職業として20年以上の経験を得た私の学びをお伝えします。
1. The document discusses using Outlook and Teams to efficiently handle multitasking. It provides tips for classifying and processing information and tasks in a timely manner.
2. It suggests using tools like Outlook, Teams, OneDrive for task management and sharing files and information. Smartphones are highly effective for initial processing of emails and tasks.
3. PCs are necessary for longer responses, document creation, coding, and research. Cloud services allow storing and accessing all work files and emails from any device.
3. DNN Processing Units
効率性柔軟性
Soft DPU
(FPGA)
Contro
l Unit
(CU)
Registers
Arithmeti
c Logic
Unit
(ALU)
CPUs GPUs
ASICsHard
DPU
Cerebras
Google TPU
Graphcore
Groq
Intel Nervana
Movidius
Wave Computing
Etc.
BrainWave
Baidu SDA
Deephi Tech
ESE
Teradeep
Etc.
5. 0 1 2
784 x 100 + 100 x 10 = 785000本
9
60000個の
教師あり学習データ
785000 x 60000 = 47100000000回の足し算をします 471億回
8. 14 days 1 hour 31 mins 15 mins
Before
2017
Apr Sept Nov
ResNet-50
NVIDIA M40 GPU
ResNet-50
32 CPU
256 Nvidia P100 GPUs
ResNet-50
1,600 CPUs
ResNet-50
1,024 P100 GPUs
Facebook
UC Berkeley, TACC,
UC Davis
Preferred Network
ChainerMN
1018 single precision operations
2017
26. • Local tools
• Local Debug
• Faster
experimentation
Single VM
Development
• Larger VMs
• GPU
Scale Up
• Multi Node
• Remote Spark
• Batch Nodes
• VM Scale Sets
Scale Out
29. Trained AI
Model
score.py
{ JSON
}schema.json conda_dependencies
.yml
Azure Machine Learning
Model Management
Run Time
Model
Registry
Image
Registry
Manifest for
Image Generation
Single
Machines
(e.g.. DSVM, IoT
Devices, local
PC)
Azure Container Service
(AKS) – Kubernetes clusters
4
1 2 3