-
PUAD: Frustratingly Simple Method for Robust Anomaly Detection
Authors:
Shota Sugawara,
Ryuji Imamura
Abstract:
Develo** an accurate and fast anomaly detection model is an important task in real-time computer vision applications. There has been much research to develop a single model that detects either structural or logical anomalies, which are inherently distinct. The majority of the existing approaches implicitly assume that the anomaly can be represented by identifying the anomalous location. However,…
▽ More
Develo** an accurate and fast anomaly detection model is an important task in real-time computer vision applications. There has been much research to develop a single model that detects either structural or logical anomalies, which are inherently distinct. The majority of the existing approaches implicitly assume that the anomaly can be represented by identifying the anomalous location. However, we argue that logical anomalies, such as the wrong number of objects, can not be well-represented by the spatial feature maps and require an alternative approach. In addition, we focused on the possibility of detecting logical anomalies by using an out-of-distribution detection approach on the feature space, which aggregates the spatial information of the feature map. As a demonstration, we propose a method that incorporates a simple out-of-distribution detection method on the feature space against state-of-the-art reconstruction-based approaches. Despite the simplicity of our proposal, our method PUAD (Picturable and Unpicturable Anomaly Detection) achieves state-of-the-art performance on the MVTec LOCO AD dataset.
△ Less
Submitted 23 February, 2024;
originally announced February 2024.
-
Exceptional point proximity-driven mode-locking in coupled microresonators
Authors:
Riku Imamura,
Shun Fujii,
Ayata Nakashima,
Takasumi Tanabe
Abstract:
We show theoretically and numerically that mode-locking is feasible with a coupled-cavity system with gain and loss, notably, without any natural saturable absorber. We highlight that in the vicinity of the exceptional point, system $Q$ exhibits substantial modulation even with minor refractive index changes and a minimal Kerr effect contribution. Leveraging this unique behaviour, we propose an un…
▽ More
We show theoretically and numerically that mode-locking is feasible with a coupled-cavity system with gain and loss, notably, without any natural saturable absorber. We highlight that in the vicinity of the exceptional point, system $Q$ exhibits substantial modulation even with minor refractive index changes and a minimal Kerr effect contribution. Leveraging this unique behaviour, we propose an unprecedented approach wherein the lossy auxiliary cavity functions as an efficient artificial saturable absorber, thus facilitating mode-locking. This approach is not only novel, but also presents considerable advantages over conventional systems where both gain and saturable absorption are contained within a single microcavity. These benefits include reduced operational power and ease of post-adjustment, achievable through the manipulation of the coupling strength between the two microcavities.
△ Less
Submitted 11 September, 2023;
originally announced September 2023.
-
Ground test results of the micro-vibration interference for the x-ray microcalorimeter onboard XRISM
Authors:
Takashi Hasebe,
Ryuta Imamura,
Masahiro Tsujimoto,
Hisamitsu Awaki,
Meng P. Chiao,
Ryuichi Fujimoto,
Leslie S. Hartz,
Caroline A. Kilbourne,
Gary A. Sneiderman,
Yoh Takei,
Susumu Yasuda
Abstract:
Resolve is a payload hosting an X-ray microcalorimeter detector operated at 50 mK in the X-Ray Imaging and Spectroscopy Mission (XRISM). It is currently under development as part of an international collaboration and is planned to be launched in 2023. A primary technical concern is the micro-vibration interference in the sensitive microcalorimeter detector caused by the spacecraft bus components.…
▽ More
Resolve is a payload hosting an X-ray microcalorimeter detector operated at 50 mK in the X-Ray Imaging and Spectroscopy Mission (XRISM). It is currently under development as part of an international collaboration and is planned to be launched in 2023. A primary technical concern is the micro-vibration interference in the sensitive microcalorimeter detector caused by the spacecraft bus components. We conducted a series of verification tests in 2021-2022 on the ground, the results of which are reported here. We defined the micro-vibration interface between the spacecraft and the Resolve instrument. In the instrument-level test, the flight-model hardware was tested against the interface level by injecting it with micro-vibrations and evaluating the instrument response using the 50 mK stage temperature stability, ADR magnet current consumption rate, and detector noise spectra. We found strong responses when injecting micro-vibration at about 200, 380, and 610 Hz. In the former two cases, the beat between the injected frequency and cryocooler frequency harmonics were observed in the detector noise spectra. In the spacecraft-level test, the acceleration and instrument responses were measured with and without suspension of the entire spacecraft. The reaction wheels (RWs) and inertial reference units (IRUs), two major sources of micro-vibration among the bus components, were operated. In conclusion, the observed responses of Resolve are within the acceptable levels in the nominal operational range of the RWs and IRUs. There is no evidence that the resultant energy
resolution degradation is beyond the current allocation of noise budget.
△ Less
Submitted 2 March, 2023;
originally announced March 2023.
-
Mechanical cryocooler noise observed in the ground testing of the Resolve X-ray microcalorimeter onboard XRISM
Authors:
R. Imamura,
H. Awaki,
M. Tsujimoto,
S. Yamada,
F. S. Porter,
C. A. Kilbourne,
R. L. Kelley,
Y. Takei
Abstract:
Low-temperature detectors often use mechanical coolers as part of the cooling chain in order to reach sub-Kelvin operating temperatures. The microphonics noise caused by the mechanical coolers is a general and inherent issue for these detectors. We have observed this effect in the ground test data obtained with the Resolve instrument to be flown on the XRISM satellite. Resolve is a cryogenic X-ray…
▽ More
Low-temperature detectors often use mechanical coolers as part of the cooling chain in order to reach sub-Kelvin operating temperatures. The microphonics noise caused by the mechanical coolers is a general and inherent issue for these detectors. We have observed this effect in the ground test data obtained with the Resolve instrument to be flown on the XRISM satellite. Resolve is a cryogenic X-ray microcalorimeter spectrometer with a required energy resolution of 7 eV at 6 keV. Five mechanical coolers are used to cool from ambient temperature to about 4 K: four two-stage Stirling coolers (STC) driven nominally at 15 Hz and a Joule-Thomson cooler (JTC) driven nominally at 52 Hz. In 2019, we operated the flight-model instrument for two weeks, in which we also obtained accelerometer data inside the cryostat at a low-temperature stage (He tank). X-ray detector and accelerometer data were obtained continuously while changing the JTC drive frequency, which produced a unique data set for investigating how the vibration from the cryocoolers propagates to the detector. In the detector noise spectra, we observed harmonics of both STCs and JTC. More interestingly, we also observed the low (<20 Hz) frequency beat between the 4'th JTC and 14'th STC harmonics and the 7'th JTC and the 23--24'th STC harmonics. We present here a description and interpretation of these measurements.
△ Less
Submitted 2 March, 2023;
originally announced March 2023.
-
Deterministic generation of a perfect soliton crystal with a saturable absorber
Authors:
Ayata Nakashima,
Shun Fujii,
Riku Imamura,
Keigo Nagashima,
Takasumi Tanabe
Abstract:
We numerically investigate the deterministic generation of a perfect soliton crystal (PSC) in an optical microresonator functionalized with a saturable absorber (SA). The SA allows the direct formation of a PSC from an initial, periodic Turing roll. It prevents passage through a chaotic state, which induces a stochastic nature as regards the number of generated dissipative Kerr solitons. We show t…
▽ More
We numerically investigate the deterministic generation of a perfect soliton crystal (PSC) in an optical microresonator functionalized with a saturable absorber (SA). The SA allows the direct formation of a PSC from an initial, periodic Turing roll. It prevents passage through a chaotic state, which induces a stochastic nature as regards the number of generated dissipative Kerr solitons. We show that PSCs form deterministically, and the number is controlled by adjusting the input power and SA parameter. Our work provides a simple approach for obtaining a stable PSC that offers an ultra-high repetition rate and a high comb output power.
△ Less
Submitted 22 December, 2021;
originally announced December 2021.
-
Expert Human-Level Driving in Gran Turismo Sport Using Deep Reinforcement Learning with Image-based Representation
Authors:
Ryuji Imamura,
Takuma Seno,
Kenta Kawamoto,
Michael Spranger
Abstract:
When humans play virtual racing games, they use visual environmental information on the game screen to understand the rules within the environments. In contrast, a state-of-the-art realistic racing game AI agent that outperforms human players does not use image-based environmental information but the compact and precise measurements provided by the environment. In this paper, a vision-based contro…
▽ More
When humans play virtual racing games, they use visual environmental information on the game screen to understand the rules within the environments. In contrast, a state-of-the-art realistic racing game AI agent that outperforms human players does not use image-based environmental information but the compact and precise measurements provided by the environment. In this paper, a vision-based control algorithm is proposed and compared with human player performances under the same conditions in realistic racing scenarios using Gran Turismo Sport (GTS), which is known as a high-fidelity realistic racing simulator. In the proposed method, the environmental information that constitutes part of the observations in conventional state-of-the-art methods is replaced with feature representations extracted from game screen images. We demonstrate that the proposed method performs expert human-level vehicle control under high-speed driving scenarios even with game screen images as high-dimensional inputs. Additionally, it outperforms the built-in AI in GTS in a time trial task, and its score places it among the top 10% approximately 28,000 human players.
△ Less
Submitted 11 November, 2021;
originally announced November 2021.
-
Design of a passively mode-locking whispering gallery mode microaser
Authors:
Tomoki S. L. Prugger Suzuki,
Ayata Nakashima,
Keigo Nagashima,
Rammaru Ishida,
Riku Imamura,
Shun Fujii,
Sze Yun Set,
Shinji Yamashita,
Takasumi Tanabe
Abstract:
Ultrahigh repetition rate lasers will become vital light sources for many future technologies; however, their realization is challenging because the cavity size must be minimized. Whispering-gallery-mode (WGM) microresonators are attractive for this purpose since they allow the strong light-matter interaction usually needed to enable mode-locking. However, the optimum parameter ranges are entirely…
▽ More
Ultrahigh repetition rate lasers will become vital light sources for many future technologies; however, their realization is challenging because the cavity size must be minimized. Whispering-gallery-mode (WGM) microresonators are attractive for this purpose since they allow the strong light-matter interaction usually needed to enable mode-locking. However, the optimum parameter ranges are entirely unknown since no experiments have yet been conducted. Here, we numerically investigate pulsed operation in a toroidal WGM microresonator with gain and saturable absorption (SA) to study the experimental feasibility. We show that dispersion is the key parameter for achieving passive mode-locking in this system. Moreover, the design guideline provided in this work can apply to any small resonators with gain and SA and is not limited to a specific cavity system.
△ Less
Submitted 13 June, 2021;
originally announced June 2021.
-
MLF-SC: Incorporating multi-layer features to sparse coding for anomaly detection
Authors:
Ryuji Imamura,
Kohei Azuma,
Atsushi Hanamoto,
Atsunori Kanemura
Abstract:
Anomalies in images occur in various scales from a small hole on a carpet to a large stain. However, anomaly detection based on sparse coding, one of the widely used anomaly detection methods, has an issue in dealing with anomalies that are out of the patch size employed to sparsely represent images. A large anomaly can be considered normal if seen in a small scale, but it is not easy to determine…
▽ More
Anomalies in images occur in various scales from a small hole on a carpet to a large stain. However, anomaly detection based on sparse coding, one of the widely used anomaly detection methods, has an issue in dealing with anomalies that are out of the patch size employed to sparsely represent images. A large anomaly can be considered normal if seen in a small scale, but it is not easy to determine a single scale (patch size) that works well for all images. Then, we propose to incorporate multi-scale features to sparse coding and improve the performance of anomaly detection. The proposed method, multi-layer feature sparse coding (MLF-SC), employs a neural network for feature extraction, and feature maps from intermediate layers of the network are given to sparse coding, whereas the standard sparse-coding-based anomaly detection method directly works on given images. We show that MLF-SC outperforms state-of-the-art anomaly detection methods including those employing deep learning. Our target data are the texture categories of the MVTec Anomaly Detection (MVTec AD) dataset, which is a modern benchmark dataset consisting of images from the real world. Our idea can be a simple and practical option to deal with practical data.
△ Less
Submitted 9 April, 2021;
originally announced April 2021.
-
Coupling of a whispering gallery mode to a silicon chip with photonic crystal
Authors:
Yuyang Zhuang,
Hajime Kumazaki,
Shun Fujii,
Riku Imamura,
Nurul Ashikin Binti Daud,
Rammaru Ishida,
Heming Chen,
Takasumi Tanabe
Abstract:
We demonstrate the efficient coupling (99.5%) of a silica whispering gallery mode microresonator directly with a silicon chip by using a silicon photonic crystal waveguide as a coupler. The efficient coupling is attributed to the small effective refractive index difference between the two devices. The large group index of the photonic crystal waveguide mode also contributes to the efficient coupli…
▽ More
We demonstrate the efficient coupling (99.5%) of a silica whispering gallery mode microresonator directly with a silicon chip by using a silicon photonic crystal waveguide as a coupler. The efficient coupling is attributed to the small effective refractive index difference between the two devices. The large group index of the photonic crystal waveguide mode also contributes to the efficient coupling. A coupling Q of 2.68*10^6 is obtained, which allows us to achieve the critical coupling of a silica whispering gallery mode with an intrinsic Q of close to 10^7 with a Si chip.
△ Less
Submitted 13 September, 2019;
originally announced September 2019.
-
Self-supervised Hyperspectral Image Restoration using Separable Image Prior
Authors:
Ryuji Imamura,
Tatsuki Itasaka,
Masahiro Okuda
Abstract:
Supervised learning with a convolutional neural network is recognized as a powerful means of image restoration. However, most such methods have been designed for application to grayscale and/or color images; therefore, they have limited success when applied to hyperspectral image restoration. This is partially owing to large datasets being difficult to collect, and also the heavy computational loa…
▽ More
Supervised learning with a convolutional neural network is recognized as a powerful means of image restoration. However, most such methods have been designed for application to grayscale and/or color images; therefore, they have limited success when applied to hyperspectral image restoration. This is partially owing to large datasets being difficult to collect, and also the heavy computational load associated with the restoration of an image with many spectral bands. To address this difficulty, we propose a novel self-supervised learning strategy for application to hyperspectral image restoration. Our method automatically creates a training dataset from a single degraded image and trains a denoising network without any clear images. Another notable feature of our method is the use of a separable convolutional layer. We undertake experiments to prove that the use of a separable network allows us to acquire the prior of a hyperspectral image and to realize efficient restoration. We demonstrate the validity of our method through extensive experiments and show that our method has better characteristics than those that are currently regarded as state-of-the-art.
△ Less
Submitted 1 July, 2019;
originally announced July 2019.
