The DENSO IT LAB Recognition and Learning Algorithm Collaborative Research Chair, established by Tokyo Tech’s School of Computing and Denso IT Laboratory, Inc. (Denso IT Lab), hosted an event at the Meeting on Image Recognition and Understanding (MIRU) 2022, a conference on image recognition held on July 25 in Himeji City, Hyogo Prefecture.
DENSO IT LAB x TOKYO TECH Discussion Night in MIRU 2022 consisted of two parts. Part 1 was a talk session by three invited speakers from Tokyo Tech and Denso IT Lab. Part 2 was a poster session involving a broad range of students, researchers, and Tokyo Tech and University of Tokyo alumni, which also served as a social gathering.
During his opening remarks, Denso IT Lab CTO Hirotoshi Iwasaki highlighted the importance of transferring basic research outcomes to real applications in a timely fashion, particularly in the fast-moving field of AI. The type of large-scale collaboration implemented by Tokyo Tech and Denso IT Lab seems to be effective in accomplishing this, Iwasaki noted.
Part 1: Invited talk session
Part 1 of the event included talks by the following participants.
Associate Professor Ikuro Sato during his talk
Ikuro Sato
Associate Professor, School of Computing, Tokyo Tech / Denso IT Lab
“Replacing Human Vision with Machines — In Search of New Designs of Deep Networks”
Rio Yokota
Associate Professor, Global Scientific Information and Computing Center, Tokyo Tech
“Large-Scale Pre-Training of Vision Transformer with Artificially Generated Images”
Shunsuke Ono
Associate Professor, School of Computing, Tokyo Tech
“Signal Processing Today — Aufheben of Model-Based and Data-Driven Approaches”
Part 2: Poster Session
Part 2 was a poster session involving the following students, researchers, and alumni.
Master’s student giving poster presentation
Tsubasa Kitayama
2nd-year master’s student, Computer Science
“Segmentation of Action Categories from Omnidirectional Videos with Temporal Extension of Spherical Convolution”
Wenru Zheng
2nd-year master’s student, Computer Science
“Event Recognition by Audio-Visual Fusion with Omnidirectional Camera and Microphone Array”
Tomoya Takahashi
2nd-year master’s student, Computer Science
“Study and Plan on Large-Scale Self-Supervised Learning of Driving Videos”
Tatsukichi Shibuya
1st-year master’s student, Computer Science
“Improving Target Propagation using Feedback Network with Random Fixed Weights”
Pablo Cervantes
3rd-year doctoral student, Computer Science
“Implicit Neural Representation Learning for Human Motion Generation (ECCV 2022)”
Toshihiro Ota
Researcher, School of Computing
“Learning with Partial Forgetting in Modern Hopfield Networks”
Tamotsu Kurioka
1st-year master’s student, Computer Science
“Study on Optimization of Data Augmentation using Multiple Teacher Models”
Ryota Yamada
2nd-year master’s student, Computer Science
“Post-Training of Feature Extractor for Improving Generalization of Deep Models (ICML 2022)”
Liang Jinrong
1sts-year master’s student, Computer Science
“Learning of Non-Uniform Step-Sizes for Neural Network Quantization”
Sora Takashima
2nd-year master’s student, Computer Science
“Replacing Labeled Real-image Datasets with Auto-generated Contours (CVPR 2022)”
Keita Takayama
Tokyo Tech alum
“Improving Generalization with Smooth Transfer Learning (NeurIPS 2021 Workshop)”
Mark Chen
University of Tokyo alum
“Improvement of View Angle Estimation from Single Images Based on Generative Features”
Aoyu Li
Tokyo Tech alum
“Informative Sample-Aware Proxy for Deep Metric Learning”
Kohta Ishikawa
Denso IT Lab
“Evaluation of Theoretical Bound of Motion Estimation Error using Delay-Aware Model for Event-Based Camera”
Shingo Yashima
Denso IT Lab
“Feature Space Particle Inference for Neural Network Ensembles (ICML 2022)”
Yuichi Yoshida
Denso IT Lab
“Resolution-Free Anomaly Detection by Implicit Image Representations”
Teppei Suzuki
Denso IT Lab
“Image Segmentation by Hierarchical Clustering using Attentions”
Yusuke Sekikawa
Denso IT Lab
“Event-Based Camera Pose Tracking via Error between Temporal Gradient of NeRF and Event”
Shin-ichi Sumiyoshi
Denso IT Lab
“Dense Shape Reconstruction using 2D Random Patterns and Motion Constraints”
Comments from participants
Thoughout the event, feedback from participants was positive. Comments included the following:
- I was very excited that the invited talks shed light on future AI research.
- The posters were all so interesting and I learned a lot from them.
- This large-scale collaboration seems to be fairly new in the information science area.
- It is intriguing to see research management for this type of academia-industry collaboration.
In his closing remarks, School of Computing Professor Koichi Shinoda, who is also the leader of the collaborative research chair, shared his thoughts. “This collaborative research chair, now consisting of nine faculty members, postdocs, many RAs from Tokyo Tech, and company researchers from Denso IT Lab, has grown into a large research group on image recognition and machine learning. You can expect more research findings to come.”
