SlideShare a Scribd company logo
Notes for
CVPR 2017: Machine Learning Sessions
Paper reviewed by Taegyun Jeon
Paper Table
Spotlight
1-1A
Exclusivity-Consistency Regularized Multi-View Subspace Clustering Xiaobo Wang et al.
Borrowing Treasures From the Wealthy: Deep Transfer Learning Through
Selective Joint Fine-Tuning
Weifeng Ge, Yizhou Yu
The More You Know: Using Knowledge Graphs for Image Classification Kenneth Marino et al.
Dynamic Edge-Conditioned Filters in Convolutional Neural Networks on Graphs Martin Simonovsky, Nikos Komodakis
Convolutional Neural Network Architecture for Geometric Matching Ignacio Rocco et al.
Deep Affordance-Grounded Sensorimotor Object Recognition Spyridon Thermos et al.
Discovering Causal Signals in Images David Lopez-Paz et al.
On Compressing Deep Models by Low Rank and Sparse Decomposition Xiyu Yu et al.
Oral 1-1A PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation Charles R. Q et al.
Universal Adversarial Perturbations Seyed-Mohsen Moosavi-Dezfooli et al.
Unsupervised Pixel-Level Domain Adaptation With Generative Adversarial Networks Konstantinos Bousmalis et al.
Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network Christian Ledig et al.
Borrowing Treasures From the Weealthy
0904 Borrowing Treasures From the Wealthy: Deep Transfer Learning Through Selective Joint Fine-Tuning
Key Idea: deep transfer learning scheme, called selective joint fine-tuning, for improving the performance of deep learning tasks with
insufficient training data.
● Shallow feature space: Gabor filters (48) + 1st and 2nd convolutional layers of AlexNet (ImageNet)
[PR12] PR-026: Notes for CVPR Machine Learning Sessions
[PR12] PR-026: Notes for CVPR Machine Learning Sessions
The More You Know
The More You Know: Using Knowledge Graphs for Image Classification
Key Idea: structured prior knowledge in the form of knowledge graphs and shows that using this knowledge improves per- formance on
image classification
(Visual Genome Graph and WordNet)
[PR12] PR-026: Notes for CVPR Machine Learning Sessions
[PR12] PR-026: Notes for CVPR Machine Learning Sessions
On Compressing Deep Models by Low Rank and
Sparse Decomposition
0928 On Compressing Deep Models by Low Rank and Sparse Decomposition
Key idea: unified framework integrating the low-rank and sparse decomposition of weight matrices with the feature map reconstructions
[PR12] PR-026: Notes for CVPR Machine Learning Sessions
[PR12] PR-026: Notes for CVPR Machine Learning Sessions
[PR12] PR-026: Notes for CVPR Machine Learning Sessions
[PR12] PR-026: Notes for CVPR Machine Learning Sessions
[PR12] PR-026: Notes for CVPR Machine Learning Sessions
[PR12] PR-026: Notes for CVPR Machine Learning Sessions
Booth Information
NVIDIA
● NVIDIA DGX-1 Station 소개
○ 가격 ~$69,000. (학교나 연구소 할인 프로모션 있음)
○ Volta 아키텍쳐 Tesla P100 x 4장 포함. (지금 DGX-1을 구매하면 pascal 아키텍쳐로 판매 이후 Volta로 업그레이드)
○ 9월경 출시 (변경가능)
○ 구매 대수에 따라 NVIDIA Cloud 플랫폼 사용권 제공
● NVIDIA Cloud
○ TensorFlow, CNTK, PyTorch, Caffe등 대부분의 모든 딥러닝 라이브러리를 NVIDIA Docker상에 제공.
○ 스케쥴링 기능 추가
○ NVIDIA DIGITS과 UI를 계승. 상당부분 개선.
● NVIDIA Jetson 보드 소개
● 학회중 Best Paper Award받은 학생들에게 젠슨황이 직접와서 GPU뿌리고 감.
● 학회에서 진행된 워크샵의 competition 입상 선물들이 대부분 NVIDIA Titan XP였음. (이번 학회의 5개 워크샵 및 튜토리얼 후원)
● NVIDIA Inception program: 스타트업들에게 플랫폼을 제공, GTC 행사에서 발표기회 제공, GPU Ventures의 투자대상 포함
● Caffe2 Meetup 행사 운영
APPLE
● 질문: MachineLearning blog 최근에 개설했는데 어떤 방향으로 진행할거냐고 물어봄
○ 대답: 계속해서 사람들을 모으고 있고 애플 제품들을 위한 서비스에 개발 (두리뭉실)
● 질문: 작년에 GAN논문 하나 내고 그뒤로 별로 paper work이 없다. 연구는 하고 있는거냐?
○ 대답: 비밀리에 하고 있다. 회사에서 내부적으로만 연구중이다.
● 지난번 NIPS와 마찬가지로 별다른 데모도 없고, 아이페드만 깔아놓고 리쿠르팅만 운영
Amazon
● Alexa, Echo등을 내새운 IoT시장을 장악하기 위한 초기 진입장벽을 허물고 있는중.
● Amazon GO등 새로운 아이템들 폭풍 선전
● Amazon A9: 아마존 온라인 플랫폼에서 상품 추천을 위해 사용되는 자체 기술. 계속 좋아지는중. 자랑자랑.
Uber
● Uber ATG, Uber Mapping에 이어 토론토에 Uber AI Lab 최근 신설
● Uber delivery, Uber X등에 사용되는 알고리즘 개발에 치중
INTEL
● Movidius Neural Compute Stick 런칭
○ 소형 플랫폼을 타겟으로 USB에 딥러닝 모델을 업로드하여 소규모 장비에도 적용 가능 ($79).
○ 불티나게 팔림.
Facebook
● 리쿠르팅: 회사 자체가 설명이 필요없는 존재. 우리 짱임. 무조건 와라. 이런 분위기.
Google
● 리쿠르팅: 회사 자체가 설명이 필요없는 존재. 우리 짱임. 무조건 와라. 이런 분위기.
Thanks
fb.com/taegyun.jeon
github.com/tgjeon
taylor.taegyun.jeon@gmail.com
Paper reviewed by
Taegyun Jeon

More Related Content

PDF
[PR12] image super resolution using deep convolutional networks
PDF
TensorFlow KR 2nd Meetup - Lightening talk (Satrec Initiative)
PDF
Deep learning super resolution
PDF
[대전AI포럼] 위성영상 분석 기술 개발 현황 소개
PPTX
Survey on Monocular Depth Estimation
PPTX
Review SRGAN
PPTX
Image retrieval
PPTX
Convolutional neural network
[PR12] image super resolution using deep convolutional networks
TensorFlow KR 2nd Meetup - Lightening talk (Satrec Initiative)
Deep learning super resolution
[대전AI포럼] 위성영상 분석 기술 개발 현황 소개
Survey on Monocular Depth Estimation
Review SRGAN
Image retrieval
Convolutional neural network

What's hot (11)

PPTX
Review EDSR
PDF
Detecting fake jpeg images
PDF
[기초개념] Graph Convolutional Network (GCN)
PPTX
Encoding in Style: a Style Encoder for Image-to-Image Translation
PDF
Deep neural networks cnn rnn_ae_some practical techniques
PDF
The bleeding edge of machine learning stream in 2017 - APAC ML/DS Community ...
PPTX
Deep Learning for AI (1)
PDF
Acceleration of Deep Neural Networks Using Stochastic Computing (확률컴퓨팅을 이용한 딥...
PDF
Workshop 210417 dhlee
PPTX
GAN with Mathematics
PDF
I3D and Kinetics datasets (Action Recognition)
Review EDSR
Detecting fake jpeg images
[기초개념] Graph Convolutional Network (GCN)
Encoding in Style: a Style Encoder for Image-to-Image Translation
Deep neural networks cnn rnn_ae_some practical techniques
The bleeding edge of machine learning stream in 2017 - APAC ML/DS Community ...
Deep Learning for AI (1)
Acceleration of Deep Neural Networks Using Stochastic Computing (확률컴퓨팅을 이용한 딥...
Workshop 210417 dhlee
GAN with Mathematics
I3D and Kinetics datasets (Action Recognition)
Ad

Similar to [PR12] PR-026: Notes for CVPR Machine Learning Sessions (20)

PDF
델켐매거진5호
PDF
[264] large scale deep-learning_on_spark
PPTX
180525 mobile visionnet_hanlim_extended
PDF
CNN Architecture A to Z
PDF
딥뉴럴넷 클러스터링 실패기
PDF
Tensorflow for Deep Learning(SK Planet)
PDF
Deep learning text NLP and Spark Collaboration . 한글 딥러닝 Text NLP & Spark
PDF
ICIP 2018 REVIEW
PPTX
Anomaly detection practive_using_deep_learning
PDF
소프트웨어 마에스트로 10기 - 책을 만나는 순간, 책을찍다
PPTX
180624 mobile visionnet_baeksucon_jwkang_pub
PPTX
[Tf2017] day4 jwkang_pub
PDF
딥러닝 논문읽기 efficient netv2 논문리뷰
PDF
기계 학습의 현재와 미래
PDF
Auto Scalable 한 Deep Learning Production 을 위한 AI Serving Infra 구성 및 AI DevOps...
PDF
Meetup history
PDF
밑바닥부터 시작하는딥러닝 8장
PDF
Summary in recent advances in deep learning for object detection
PDF
Summary in recent advances in deep learning for object detection
PDF
Feature Pyramid Network, FPN
델켐매거진5호
[264] large scale deep-learning_on_spark
180525 mobile visionnet_hanlim_extended
CNN Architecture A to Z
딥뉴럴넷 클러스터링 실패기
Tensorflow for Deep Learning(SK Planet)
Deep learning text NLP and Spark Collaboration . 한글 딥러닝 Text NLP & Spark
ICIP 2018 REVIEW
Anomaly detection practive_using_deep_learning
소프트웨어 마에스트로 10기 - 책을 만나는 순간, 책을찍다
180624 mobile visionnet_baeksucon_jwkang_pub
[Tf2017] day4 jwkang_pub
딥러닝 논문읽기 efficient netv2 논문리뷰
기계 학습의 현재와 미래
Auto Scalable 한 Deep Learning Production 을 위한 AI Serving Infra 구성 및 AI DevOps...
Meetup history
밑바닥부터 시작하는딥러닝 8장
Summary in recent advances in deep learning for object detection
Summary in recent advances in deep learning for object detection
Feature Pyramid Network, FPN
Ad

More from Taegyun Jeon (12)

PDF
TensorFlow-KR 3rd meetup - Lightning Talk for SI Analytics
PDF
TensorFlow Dev Summit 2018 Extended: TensorFlow Eager Execution
PDF
[OSGeo-KR Tech Workshop] Deep Learning for Single Image Super-Resolution
PDF
[PR12] PR-063: Peephole predicting network performance before training
PDF
GDG DevFest Xiamen 2017
PPTX
[PR12] PR-050: Convolutional LSTM Network: A Machine Learning Approach for Pr...
PDF
GDG DevFest Seoul 2017: Codelab - Time Series Analysis for Kaggle using Tenso...
PDF
[PR12] PR-036 Learning to Remember Rare Events
PDF
[PR12] You Only Look Once (YOLO): Unified Real-Time Object Detection
PDF
Google Dev Summit Extended Seoul - TensorFlow: Tensorboard & Keras
PPTX
인공지능: 변화와 능력개발
PPTX
Electricity price forecasting with Recurrent Neural Networks
TensorFlow-KR 3rd meetup - Lightning Talk for SI Analytics
TensorFlow Dev Summit 2018 Extended: TensorFlow Eager Execution
[OSGeo-KR Tech Workshop] Deep Learning for Single Image Super-Resolution
[PR12] PR-063: Peephole predicting network performance before training
GDG DevFest Xiamen 2017
[PR12] PR-050: Convolutional LSTM Network: A Machine Learning Approach for Pr...
GDG DevFest Seoul 2017: Codelab - Time Series Analysis for Kaggle using Tenso...
[PR12] PR-036 Learning to Remember Rare Events
[PR12] You Only Look Once (YOLO): Unified Real-Time Object Detection
Google Dev Summit Extended Seoul - TensorFlow: Tensorboard & Keras
인공지능: 변화와 능력개발
Electricity price forecasting with Recurrent Neural Networks

[PR12] PR-026: Notes for CVPR Machine Learning Sessions

  • 1. Notes for CVPR 2017: Machine Learning Sessions Paper reviewed by Taegyun Jeon
  • 2. Paper Table Spotlight 1-1A Exclusivity-Consistency Regularized Multi-View Subspace Clustering Xiaobo Wang et al. Borrowing Treasures From the Wealthy: Deep Transfer Learning Through Selective Joint Fine-Tuning Weifeng Ge, Yizhou Yu The More You Know: Using Knowledge Graphs for Image Classification Kenneth Marino et al. Dynamic Edge-Conditioned Filters in Convolutional Neural Networks on Graphs Martin Simonovsky, Nikos Komodakis Convolutional Neural Network Architecture for Geometric Matching Ignacio Rocco et al. Deep Affordance-Grounded Sensorimotor Object Recognition Spyridon Thermos et al. Discovering Causal Signals in Images David Lopez-Paz et al. On Compressing Deep Models by Low Rank and Sparse Decomposition Xiyu Yu et al. Oral 1-1A PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation Charles R. Q et al. Universal Adversarial Perturbations Seyed-Mohsen Moosavi-Dezfooli et al. Unsupervised Pixel-Level Domain Adaptation With Generative Adversarial Networks Konstantinos Bousmalis et al. Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network Christian Ledig et al.
  • 3. Borrowing Treasures From the Weealthy 0904 Borrowing Treasures From the Wealthy: Deep Transfer Learning Through Selective Joint Fine-Tuning Key Idea: deep transfer learning scheme, called selective joint fine-tuning, for improving the performance of deep learning tasks with insufficient training data. ● Shallow feature space: Gabor filters (48) + 1st and 2nd convolutional layers of AlexNet (ImageNet)
  • 6. The More You Know The More You Know: Using Knowledge Graphs for Image Classification Key Idea: structured prior knowledge in the form of knowledge graphs and shows that using this knowledge improves per- formance on image classification (Visual Genome Graph and WordNet)
  • 9. On Compressing Deep Models by Low Rank and Sparse Decomposition 0928 On Compressing Deep Models by Low Rank and Sparse Decomposition Key idea: unified framework integrating the low-rank and sparse decomposition of weight matrices with the feature map reconstructions
  • 16. Booth Information NVIDIA ● NVIDIA DGX-1 Station 소개 ○ 가격 ~$69,000. (학교나 연구소 할인 프로모션 있음) ○ Volta 아키텍쳐 Tesla P100 x 4장 포함. (지금 DGX-1을 구매하면 pascal 아키텍쳐로 판매 이후 Volta로 업그레이드) ○ 9월경 출시 (변경가능) ○ 구매 대수에 따라 NVIDIA Cloud 플랫폼 사용권 제공 ● NVIDIA Cloud ○ TensorFlow, CNTK, PyTorch, Caffe등 대부분의 모든 딥러닝 라이브러리를 NVIDIA Docker상에 제공. ○ 스케쥴링 기능 추가 ○ NVIDIA DIGITS과 UI를 계승. 상당부분 개선. ● NVIDIA Jetson 보드 소개 ● 학회중 Best Paper Award받은 학생들에게 젠슨황이 직접와서 GPU뿌리고 감. ● 학회에서 진행된 워크샵의 competition 입상 선물들이 대부분 NVIDIA Titan XP였음. (이번 학회의 5개 워크샵 및 튜토리얼 후원) ● NVIDIA Inception program: 스타트업들에게 플랫폼을 제공, GTC 행사에서 발표기회 제공, GPU Ventures의 투자대상 포함 ● Caffe2 Meetup 행사 운영
  • 17. APPLE ● 질문: MachineLearning blog 최근에 개설했는데 어떤 방향으로 진행할거냐고 물어봄 ○ 대답: 계속해서 사람들을 모으고 있고 애플 제품들을 위한 서비스에 개발 (두리뭉실) ● 질문: 작년에 GAN논문 하나 내고 그뒤로 별로 paper work이 없다. 연구는 하고 있는거냐? ○ 대답: 비밀리에 하고 있다. 회사에서 내부적으로만 연구중이다. ● 지난번 NIPS와 마찬가지로 별다른 데모도 없고, 아이페드만 깔아놓고 리쿠르팅만 운영 Amazon ● Alexa, Echo등을 내새운 IoT시장을 장악하기 위한 초기 진입장벽을 허물고 있는중. ● Amazon GO등 새로운 아이템들 폭풍 선전 ● Amazon A9: 아마존 온라인 플랫폼에서 상품 추천을 위해 사용되는 자체 기술. 계속 좋아지는중. 자랑자랑.
  • 18. Uber ● Uber ATG, Uber Mapping에 이어 토론토에 Uber AI Lab 최근 신설 ● Uber delivery, Uber X등에 사용되는 알고리즘 개발에 치중 INTEL ● Movidius Neural Compute Stick 런칭 ○ 소형 플랫폼을 타겟으로 USB에 딥러닝 모델을 업로드하여 소규모 장비에도 적용 가능 ($79). ○ 불티나게 팔림. Facebook ● 리쿠르팅: 회사 자체가 설명이 필요없는 존재. 우리 짱임. 무조건 와라. 이런 분위기. Google ● 리쿠르팅: 회사 자체가 설명이 필요없는 존재. 우리 짱임. 무조건 와라. 이런 분위기.