Yamato OKAMOTO

↑
ICLR’2020
COVID-19 Virtual Conference
• MTG
• MTG
•
MTG


Virtual Conference
• 100$
•
•
• WiFi
•
•
•
memo
• Disentangle
• !?
•
Virtual
Contrastive Learning of Structured World Models
•
https://openreview.net/pdf?id=H1gax6VtDB
• State Encoder Action Transition
• State+Action State Encoder 

Reconstruction State
•
Doing for Our Robots What Nature Did For Us
Prof. Leslie Kaelbling / MIT
•
https://iclr.cc/virtual/speaker_2.html#stream
Evaluating The Search Phase of Neural Architecture Search
• NAS
•
Policy
• 

(i) a constrained search space 

(ii) weight sharing, which shuffles the architecture ranking during the search, thus negatively impacting it.
https://openreview.net/pdf?id=H1loF2NFwr
Measuring and Improving the Use of Graph Information
in Graph Neural Networks
• Graph-CNN Graph-Data
• Graph-Data
•
• GNN
• Information-Gain=0 KL
https://openreview.net/pdf?id=rkeIIkHKvS
Bounds on Over-Parameterization for Guaranteed Existence of
Descent Paths in Shallow ReLU Networks
• Over-Parameter Over-Parameter
• ReLU
•
https://openreview.net/pdf?id=BkgXHTNtvS
Federated Adversarial Domain Adaptation
• Federated-Learning Source-Data
Target-Data Unsupervised-Domain-Adaptation
•
• aggregation
https://arxiv.org/pdf/1911.02054.pdf
Demystifying Inter-Class Disentanglement
• Disentangle
• Encoder Class Content
• Encoder
• VAE
https://openreview.net/pdf?id=Hyl9xxHYPr
Theory and Evaluation Metrics for Learning Disentangled
Representations
• Semantic Feature Disentangle Disentangle
•
• Disentangle
•
https://arxiv.org/pdf/1908.09961.pdf

ICLR'2020 参加速報

  • 1.
  • 2.
    COVID-19 Virtual Conference •MTG • MTG • MTG 

  • 3.
    Virtual Conference • 100$ • • •WiFi • • • memo • Disentangle • !? • Virtual
  • 5.
    Contrastive Learning ofStructured World Models • https://openreview.net/pdf?id=H1gax6VtDB • State Encoder Action Transition • State+Action State Encoder 
 Reconstruction State •
  • 6.
    Doing for OurRobots What Nature Did For Us Prof. Leslie Kaelbling / MIT • https://iclr.cc/virtual/speaker_2.html#stream
  • 7.
    Evaluating The SearchPhase of Neural Architecture Search • NAS • Policy • 
 (i) a constrained search space 
 (ii) weight sharing, which shuffles the architecture ranking during the search, thus negatively impacting it. https://openreview.net/pdf?id=H1loF2NFwr
  • 8.
    Measuring and Improvingthe Use of Graph Information in Graph Neural Networks • Graph-CNN Graph-Data • Graph-Data • • GNN • Information-Gain=0 KL https://openreview.net/pdf?id=rkeIIkHKvS
  • 9.
    Bounds on Over-Parameterizationfor Guaranteed Existence of Descent Paths in Shallow ReLU Networks • Over-Parameter Over-Parameter • ReLU • https://openreview.net/pdf?id=BkgXHTNtvS
  • 10.
    Federated Adversarial DomainAdaptation • Federated-Learning Source-Data Target-Data Unsupervised-Domain-Adaptation • • aggregation https://arxiv.org/pdf/1911.02054.pdf
  • 11.
    Demystifying Inter-Class Disentanglement •Disentangle • Encoder Class Content • Encoder • VAE https://openreview.net/pdf?id=Hyl9xxHYPr
  • 12.
    Theory and EvaluationMetrics for Learning Disentangled Representations • Semantic Feature Disentangle Disentangle • • Disentangle • https://arxiv.org/pdf/1908.09961.pdf