The University of Tokyo · 컴퓨터과학
Katsuma Inoue 교수의 연구실은 생체와 유사한 연성 물체와 비정형 컴퓨팅을 기반으로 한 혁신적 시스템 설계를 연구합니다. 소프트 컨티누üm 바디의 비틀림과 변형을 비모수적 방식으로 추론하고, 물리적 신경망을 구현하기 위한 생물학적으로 영감을 받은 학습 알고리즘을 개발합니다. 특히, 신경망의 역전파를 대체할 수 있는 직접 피드백 정렬 기반의 물리 기반 딥러닝 및 혼돈적 이탈성(chaotic itinerancy)을 활용한 자율 행동 제어 기법을 중심으로 연구를 전개하고 있습니다.
표시된 성과는 수집된 데이터 기준으로 산출되며, 일부 차이가 있을 수 있습니다.
Ever-growing demand for artificial intelligence has motivated research on unconventional computation based on physical devices. While such computation devices mimic brain-inspired analog information processing, the learning procedures still rely on methods optimized for digital processing such as backpropagation, which is not suitable for physical implementation. Here, we present physical deep learning by extending a biologically inspired training algorithm called direct feedback alignment. Unli
Soft continuum bodies have demonstrated their effectiveness in generating flexible and adaptive functionalities by capitalizing on the rich deformability of soft material. Compared with a rigid-body robot, it is in general difficult to model and emulate the morphology dynamics of a soft continuum body. In addition, a soft continuum body potentially has an infinite degree of freedom, requiring considerable labor to manually annotate its dynamics from external sensory data such as video. In this s
Language is an outcome of our complex and dynamic human-interactions and the technique of natural language processing (NLP) is hence built on human linguistic activities. Along with generative pretrained transformer (GPT), bidirectional encoder representations from transformers (bert) has recently gained its popularity, owing to its outstanding NLP capabilities, by establishing the state-of-the-art scores in several NLP benchmarks. A lite bert (albert) is literally characterized as a lightweight
Biological systems are composed of a large number of continuum elements and they effectively realize their flexible and smooth functionalities by capitalizing on the rich dynamics occurring in their soft bodies. In particular, the diverse spatiotemporal patterns of the continuum body have recently been demonstrated as highly useful in implementing a certain class of computation, implying that soft bodies can work as computational devices like nervous systems. However, unlike nervous systems whos
Language is an outcome of our complex and dynamic human-interactions and the technique of natural language processing (NLP) is hence built on human linguistic activities. Bidirectional Encoder Representations from Transformers (BERT) has recently gained its popularity by establishing the state-of-the-art scores in several NLP benchmarks. A Lite BERT (ALBERT) is literally characterized as a lightweight version of BERT, in which the number of BERT parameters is reduced by repeatedly applying the s
Chaotic itinerancy is a frequently observed phenomenon in high-dimensional and nonlinear dynamical systems, and it is characterized by the random transitions among multiple quasi-attractors. Several studies have revealed that chaotic itinerancy has been observed in brain activity, and it is considered to play a critical role in the spontaneous, stable behavior generation of animals. Thus, chaotic itinerancy is a topic of great interest, particularly for neurorobotics researchers who wish to unde
Reservoir computing (RC) is a machine learning framework that uses recurrent neural networks and is characterized by directly capitalizing on intrinsic dynamics instead of adjusting internal parameters. In particular, in the form of physical reservoir computing (PRC), recent studies have advanced by treating various physical systems as reservoirs and applying them to time-series data processing and quantifying information-processing properties. In this way, RC and PRC potentially have interdisci
glossaries
glossaries
乳癌検診は，検診を行う地域の地理的条件や読影医師数などによりその運営に幅があり，全国レベルで精度を底上げする方法の開発が待たれる。今回われわれは，マンモグラフィの読影判定を，ディープラーニングを用いて自動的に判定することでその精度を計測し，検診に応用できるか検討した。乳癌の存在が確認されているマンモグラフィ１０４症例を正方形の画像２０枚に自動的に切り出し，反転もさせた合計２，０４８枚の画像に対し畳み込みニューラルネットワークを用いて判定，乳癌を含む画像かどうかを判断させ，その正診率を計測した。学習させた結果，正診率は９４．９％，感度は８８．５％，特異度は９７．１％，陽性適中率は９１．１％，陰性適中率は９６．１％となった。高い正診率を誇るディープラーニングを乳癌検診に用いることで，将来乳癌検診の全国規模での精度向上や効率化に寄与する可能性があると思われた。
glossaries
We demonstrate back-propagation-free photonic deep learning by extending a biologically inspired training called direct feedback alignment. The potential for accelerated computation with competitive performance was confirmed through numerical simulation and optoelectric hardware implementation.
glossaries
