【DL輪読会】Efficiently Modeling Long Sequences with Structured State SpacesDeep Learning JP
This document summarizes a research paper on modeling long-range dependencies in sequence data using structured state space models and deep learning. The proposed S4 model (1) derives recurrent and convolutional representations of state space models, (2) improves long-term memory using HiPPO matrices, and (3) efficiently computes state space model convolution kernels. Experiments show S4 outperforms existing methods on various long-range dependency tasks, achieves fast and memory-efficient computation comparable to efficient Transformers, and performs competitively as a general sequence model.
This document provides an overview of POMDP (Partially Observable Markov Decision Process) and its applications. It first defines the key concepts of POMDP such as states, actions, observations, and belief states. It then uses the classic Tiger problem as an example to illustrate these concepts. The document discusses different approaches to solve POMDP problems, including model-based methods that learn the environment model from data and model-free reinforcement learning methods. Finally, it provides examples of applying POMDP to games like ViZDoom and robot navigation problems.
Several recent papers have explored self-supervised learning methods for vision transformers (ViT). Key approaches include:
1. Masked prediction tasks that predict masked patches of the input image.
2. Contrastive learning using techniques like MoCo to learn representations by contrasting augmented views of the same image.
3. Self-distillation methods like DINO that distill a teacher ViT into a student ViT using different views of the same image.
4. Hybrid approaches that combine masked prediction with self-distillation, such as iBOT.
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.
【DL輪読会】Efficiently Modeling Long Sequences with Structured State SpacesDeep Learning JP
This document summarizes a research paper on modeling long-range dependencies in sequence data using structured state space models and deep learning. The proposed S4 model (1) derives recurrent and convolutional representations of state space models, (2) improves long-term memory using HiPPO matrices, and (3) efficiently computes state space model convolution kernels. Experiments show S4 outperforms existing methods on various long-range dependency tasks, achieves fast and memory-efficient computation comparable to efficient Transformers, and performs competitively as a general sequence model.
This document provides an overview of POMDP (Partially Observable Markov Decision Process) and its applications. It first defines the key concepts of POMDP such as states, actions, observations, and belief states. It then uses the classic Tiger problem as an example to illustrate these concepts. The document discusses different approaches to solve POMDP problems, including model-based methods that learn the environment model from data and model-free reinforcement learning methods. Finally, it provides examples of applying POMDP to games like ViZDoom and robot navigation problems.
Several recent papers have explored self-supervised learning methods for vision transformers (ViT). Key approaches include:
1. Masked prediction tasks that predict masked patches of the input image.
2. Contrastive learning using techniques like MoCo to learn representations by contrasting augmented views of the same image.
3. Self-distillation methods like DINO that distill a teacher ViT into a student ViT using different views of the same image.
4. Hybrid approaches that combine masked prediction with self-distillation, such as iBOT.
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.
SSII2021 [SS2] Deepfake Generation and Detection – An Overview (ディープフェイクの生成と検出)SSII
This document provides an overview of deepfake generation and detection. It begins with an introduction to the author and their background and research interests. The rest of the document is outlined as follows: definitions of deepfakes, various deepfake generation techniques including face synthesis, manipulation, reenactment and swapping, and an overview of deepfake detection methods including commonly used datasets, image-based and video-based detection approaches.
10. 本日の発表内容
1. “Deployment-efficiency” in learning controls (algorithm view)
● Show model-learning can benefit in reinforcement learning from offline
data
● Propose an algorithm interleaving model/policy learning and batched data
collection
2. Development of real service robot systems (system view)
● Building baseline systems of service robots in the house
● Study how learning modules can be integrated into robot systems
10
11. Deployment-Efficient Reinforcement Learning
via Model-Based Offline Optimization
Tatsuya Matsushima1
*, Hiroki Furuta1
*, Yutaka Matsuo1
,
Ofir Nachum2
, Shixiang Shane Gu2
1
The University of Tokyo, 2
Google Brain (*Contributed Equally)
Contact: matsushima@weblab.t.u-tokyo.ac.jp
ICLR2021
30. Development of Partner Robot System
Using Toyota HSR
Team Weblab
Team Leader: Tatsuya Matsushima
Advisor: Yusuke Iwasawa & Yutaka Matsuo
Team Contact: robocup@weblab.t.u-tokyo.ac.jp
45. Simulator-to-Real (Sim2Real) Transfer in Recognition
45
Sim2Real of grasp pose prediction
● Generated dataset with simulator (PyBullet) with randomized objects
○ Using ShapeNet Objects (7000+) instead of YCB objects (70+)
● Learn FCN model with depth image (grasp pose regression)