Naoto Sato
Research Subjects
Research Interests
Naoto Sato has a total of 51 co-authors. Here are the top co-authors by number of shared publications:
Hideto Ogawa
Research & Development Group Hitachi, Ltd
3 shared publicationsHironobu Kuruma
Research & Development Group Hitachi, Ltd
3 shared publications- Mithun Kuniyil Ajith Singh
CYBERDYNE Inc
3 shared publications Fumiyuki Ichihashi
Cyberdyne, Inc. (Japan)
2 shared publicationsRyota Katsube
Katsube are with Research & Development Group Hitachi, Ltd Japan
2 shared publications- Thai Son Hoang
University of Southampton
2 shared publications
Latest Publications
- Open Access
- 23 September 2025
- Published by Elsevier in Science of Remote Sensing
- Vol. 12
Improving soil moisture estimation in wet soils using L-band Synthetic Aperture Radar (SAR) through polarization and filtering optimization
- 23 September 2025
- Published by Elsevier in Science of Remote Sensing
- Vol. 12
Open AccessImproving soil moisture estimation in wet soils using L-band Synthetic Aperture Radar (SAR) through polarization and filtering optimization
- Open Access
Bug-locating Method based on Statistical Testing for Quantum Programs
When a bug is detected by testing a quantum program on a quantum computer, we want to determine its location to fix it. To locate the bug, the quantum program is divided into several segments, and each segment is tested. However, to prepare a quantum state that is input to a segment, it is necessary to execute all the segments ahead of that segment in a quantum computer. This means that the cost of testing each segment depends on its location. We can also locate a buggy segment only if it is confirmed that there are no bugs in all segments ahead of that buggy segment. Since a quantum program is tested statistically on the basis of measurement results, there is a tradeoff between testing accuracy and cost. These characteristics are unique to quantum programs and complicate locating bugs. We propose an efficient bug-locating method consisting of four approaches, cost-based binary search, early determination, finalization, and looking back, which take these characteristics into account. We present experimental results that indicate that the proposed method can reduce the bug-locating cost, represented as the number of executed quantum gates, compared to naive methods that do not use the four approaches. The limitation and usefulness of the proposed method are also discussed from the experimental results.
Open AccessBug-locating Method based on Statistical Testing for Quantum Programs
When a bug is detected by testing a quantum program on a quantum computer, we want to determine its location to fix it. To locate the bug, the quantum program is divided into several segments, and each segment is tested. However, to prepare a quantum state that is input to a segment, it is necessary to execute all the segments ahead of that segment in a quantum computer. This means that the cost of testing each segment depends on its location. We can also locate a buggy segment only if it is confirmed that there are no bugs in all segments ahead of that buggy segment. Since a quantum program is tested statistically on the basis of measurement results, there is a tradeoff between testing accuracy and cost. These characteristics are unique to quantum programs and complicate locating bugs. We propose an efficient bug-locating method consisting of four approaches, cost-based binary search, early determination, finalization, and looking back, which take these characteristics into account. We present experimental results that indicate that the proposed method can reduce the bug-locating cost, represented as the number of executed quantum gates, compared to naive methods that do not use the four approaches. The limitation and usefulness of the proposed method are also discussed from the experimental results.
- Open Access
Fast Evaluation of DNN for Past Dataset in Incremental Learning
During the operation of a system including a deep neural network (DNN), new input values that were not included in the training dataset are given to the DNN. In such a case, the DNN may be incrementally trained with the new input values; however, that training may reduce the accuracy of the DNN in regard to the dataset that was previously obtained and used for the past training. It is necessary to evaluate the effect of the additional training on the accuracy for the past dataset. However, evaluation by testing all the input values included in the past dataset takes time. Therefore, we propose a new method to quickly evaluate the effect on the accuracy for the past dataset. In the proposed method, the gradient of the parameter values (such as weight and bias) for the past dataset is extracted by running the DNN before the training. Then, after the training, its effect on the accuracy with respect to the past dataset is calculated from the gradient and update differences of the parameter values. To show the usefulness of the proposed method, we present experimental results with several datasets. The results show that the proposed method can estimate the accuracy change by additional training in a constant time.
Open AccessFast Evaluation of DNN for Past Dataset in Incremental Learning
During the operation of a system including a deep neural network (DNN), new input values that were not included in the training dataset are given to the DNN. In such a case, the DNN may be incrementally trained with the new input values; however, that training may reduce the accuracy of the DNN in regard to the dataset that was previously obtained and used for the past training. It is necessary to evaluate the effect of the additional training on the accuracy for the past dataset. However, evaluation by testing all the input values included in the past dataset takes time. Therefore, we propose a new method to quickly evaluate the effect on the accuracy for the past dataset. In the proposed method, the gradient of the parameter values (such as weight and bias) for the past dataset is extracted by running the DNN before the training. Then, after the training, its effect on the accuracy with respect to the past dataset is calculated from the gradient and update differences of the parameter values. To show the usefulness of the proposed method, we present experimental results with several datasets. The results show that the proposed method can estimate the accuracy change by additional training in a constant time.
- Open Access
Generating Shuttling Procedures for Constrained Silicon Quantum Dot Array
In silicon quantum computers, a single electron is trapped in a microstructure called a quantum dot, and its spin is used as a qubit. For large-scale integration of qubits, we previously proposed an approach of arranging the quantum dots in a two-dimensional array and sharing a control gate in a row or column of the array. In our array, the shuttling of electrons is a useful technique to operate the target qubit independently and avoid crosstalk. However, since the shuttling is also conducted using shared control gates, the movement of qubits is complexly constrained. We therefore propose a formal model on the basis of state transition systems to describe those constraints and operation procedures on the array. We also present an approach to generate operation procedures under the constraints. Utilizing this approach, we present a concrete method for our 16 $\times$ 8 quantum dot array. By implementing the proposed method as a quantum compiler, we confirmed that it is possible to generate operation procedures in a practical amount of time for arbitrary quantum circuits. We also demonstrated that crosstalk can be avoided by shuttling and that the fidelity in that case is higher than when crosstalk is not avoided.Open AccessGenerating Shuttling Procedures for Constrained Silicon Quantum Dot Array
In silicon quantum computers, a single electron is trapped in a microstructure called a quantum dot, and its spin is used as a qubit. For large-scale integration of qubits, we previously proposed an approach of arranging the quantum dots in a two-dimensional array and sharing a control gate in a row or column of the array. In our array, the shuttling of electrons is a useful technique to operate the target qubit independently and avoid crosstalk. However, since the shuttling is also conducted using shared control gates, the movement of qubits is complexly constrained. We therefore propose a formal model on the basis of state transition systems to describe those constraints and operation procedures on the array. We also present an approach to generate operation procedures under the constraints. Utilizing this approach, we present a concrete method for our 16 $\times$ 8 quantum dot array. By implementing the proposed method as a quantum compiler, we confirmed that it is possible to generate operation procedures in a practical amount of time for arbitrary quantum circuits. We also demonstrated that crosstalk can be avoided by shuttling and that the fidelity in that case is higher than when crosstalk is not avoided. - Open Access
Locating Buggy Segments in Quantum Program Debugging
When a bug is detected by testing a quantum program on a quantum computer, we want to determine its location to fix it. To locate the bug, the quantum program is divided into several segments, and each segment is tested. However, to prepare a quantum state that is input to a segment, it is necessary to execute all the segments ahead of that segment in a quantum computer. This means that the cost of testing each segment depends on its location. We can also locate a buggy segment only if it is confirmed that there are no bugs in all segments ahead of that buggy segment. Since a quantum program is tested statistically on the basis of measurement results, there is a tradeoff between testing accuracy and cost. Although these characteristics are unique to quantum programs and complicate locating bugs, they have not been investigated. We suggest for the first time that these characteristics should be considered to efficiently locate bugs. We are also the first to propose a bug-locating method that takes these characteristics into account. The results from experiments indicate that the bug-locating cost, represented as the number of executed quantum gates, can be reduced with the proposed method compared with naive methods.
Open AccessLocating Buggy Segments in Quantum Program Debugging
When a bug is detected by testing a quantum program on a quantum computer, we want to determine its location to fix it. To locate the bug, the quantum program is divided into several segments, and each segment is tested. However, to prepare a quantum state that is input to a segment, it is necessary to execute all the segments ahead of that segment in a quantum computer. This means that the cost of testing each segment depends on its location. We can also locate a buggy segment only if it is confirmed that there are no bugs in all segments ahead of that buggy segment. Since a quantum program is tested statistically on the basis of measurement results, there is a tradeoff between testing accuracy and cost. Although these characteristics are unique to quantum programs and complicate locating bugs, they have not been investigated. We suggest for the first time that these characteristics should be considered to efficiently locate bugs. We are also the first to propose a bug-locating method that takes these characteristics into account. The results from experiments indicate that the bug-locating cost, represented as the number of executed quantum gates, can be reduced with the proposed method compared with naive methods.
- 16 May 2021
- Published by Springer Nature in Clinical Journal of Gastroenterology
- Vol. 14 (4) , 1-6
Successful control of portal hypertension-related complications after two embolization procedures for multiple and large spontaneous portosystemic shunts in a patient with liver cirrhosis
Liver cirrhosis is frequently complicated by spontaneous portosystemic shunt (SPSS) due to portal hypertension. Shunt embolization is considered when symptoms related to SPSSs are refractory to endoscopic and/or medical therapies. However, little information is available on the treatment of patients with multiple and large SPSS. We report a successfully managed case in which patient with such SPSS received two embolization procedures within 6 months. A 57-year-old man with alcoholic liver cirrhosis was transferred to our hospital due to a ruptured gastric varix. CT examination showed gastrorenal and splenorenal shunts of 8 mm and 11 mm in diameter, respectively. In addition, multiple hepatocellular carcinomas (HCCs) were noted. First, balloon-occluded retrograde transvenous obliteration (BRTO) was performed for the gastrorenal shunt, resulting in the disappearance of the varix, followed by transcatheter arterial chemoembolization (TACE) for HCCs. However, the hepatic encephalopathy worsened after the BRTO and TACE, and the splenorenal shunt enlarged to 18 mm in diameter. Although the shunt was tortuous and had another drainage vein, we completed the embolization for the shunt using metallic coils without any events. The patient’s hepatic encephalopathy and hepatic function were ameliorated after embolization for the splenorenal shunt, and the patient was free from hepatic encephalopathy.
- 16 May 2021
- Published by Springer Nature in Clinical Journal of Gastroenterology
- Vol. 14 (4) , 1-6
Successful control of portal hypertension-related complications after two embolization procedures for multiple and large spontaneous portosystemic shunts in a patient with liver cirrhosis
Liver cirrhosis is frequently complicated by spontaneous portosystemic shunt (SPSS) due to portal hypertension. Shunt embolization is considered when symptoms related to SPSSs are refractory to endoscopic and/or medical therapies. However, little information is available on the treatment of patients with multiple and large SPSS. We report a successfully managed case in which patient with such SPSS received two embolization procedures within 6 months. A 57-year-old man with alcoholic liver cirrhosis was transferred to our hospital due to a ruptured gastric varix. CT examination showed gastrorenal and splenorenal shunts of 8 mm and 11 mm in diameter, respectively. In addition, multiple hepatocellular carcinomas (HCCs) were noted. First, balloon-occluded retrograde transvenous obliteration (BRTO) was performed for the gastrorenal shunt, resulting in the disappearance of the varix, followed by transcatheter arterial chemoembolization (TACE) for HCCs. However, the hepatic encephalopathy worsened after the BRTO and TACE, and the splenorenal shunt enlarged to 18 mm in diameter. Although the shunt was tortuous and had another drainage vein, we completed the embolization for the shunt using metallic coils without any events. The patient’s hepatic encephalopathy and hepatic function were ameliorated after embolization for the splenorenal shunt, and the patient was free from hepatic encephalopathy.
- 5 March 2021
- Published by SPIE-Intl Soc Optical Eng
- p. 88
Optimizing acoustic detection for deep-tissue LED-based photoacoustic imaging
SPIE Digital Library Proceedings
- 5 March 2021
- Published by SPIE-Intl Soc Optical Eng
- p. 88
Optimizing acoustic detection for deep-tissue LED-based photoacoustic imaging
SPIE Digital Library Proceedings
- Open Access
- 31 December 2020
- Published by Center for Academic Publications Japan in Journal of Nutritional Science and Vitaminology
- Vol. 66 (6) , 587-590
Mastication of Hard Gumi Decreases the Gustatory Threshold for Sodium Chloride
In recent years, there has been an increase in the number of hypertensive diseases and the various diseases associated with them. A major cause of these is excessive salt intake. The purpose of the present study was to examine whether chewing hard foods lowers the saltiness threshold. Fifteen subjects (fourteen women and one man) participated in the present study. Two types of gummies are available as ingredients: hard and soft gummies. The saltiness thresholds before and after chewing of each gummi were studied using 11 different NaCl solutions. Then, points of subjective equality (PSEs) were calculated to detect changes in the saltiness for each subject. In the soft Gumi condition, there was no significant difference in PSE for the saltiness between before and after ingesting Gumi (p>0.05), while in the hard Gumi condition, the PSE for the saltiness significantly decreased after ingesting Gumi compared with the value of before ingesting Gumi (p=0.001). From these results, we concluded that sensitivity to saltiness would increase after mastication of hard foods such as hard Gumi.
- 31 December 2020
- Published by Center for Academic Publications Japan in Journal of Nutritional Science and Vitaminology
- Vol. 66 (6) , 587-590
Open AccessMastication of Hard Gumi Decreases the Gustatory Threshold for Sodium Chloride
In recent years, there has been an increase in the number of hypertensive diseases and the various diseases associated with them. A major cause of these is excessive salt intake. The purpose of the present study was to examine whether chewing hard foods lowers the saltiness threshold. Fifteen subjects (fourteen women and one man) participated in the present study. Two types of gummies are available as ingredients: hard and soft gummies. The saltiness thresholds before and after chewing of each gummi were studied using 11 different NaCl solutions. Then, points of subjective equality (PSEs) were calculated to detect changes in the saltiness for each subject. In the soft Gumi condition, there was no significant difference in PSE for the saltiness between before and after ingesting Gumi (p>0.05), while in the hard Gumi condition, the PSE for the saltiness significantly decreased after ingesting Gumi compared with the value of before ingesting Gumi (p=0.001). From these results, we concluded that sensitivity to saltiness would increase after mastication of hard foods such as hard Gumi.
- Open Access
Unsupposable Test-data Generation for Machine-learned Software
As for software development by machine learning, a trained model is evaluated by using part of an existing dataset as test data. However, if data with characteristics that differ from the existing data is input, the model does not always behave as expected. Accordingly, to confirm the behavior of the model more strictly, it is necessary to create data that differs from the existing data and test the model with that different data. The data to be tested includes not only data that developers can suppose (supposable data) but also data they cannot suppose (unsupposable data). To confirm the behavior of the model strictly, it is important to create as much unsupposable data as possible. In this study, therefore, a method called "unsupposable test-data generation" (UTG)---for giving suggestions for unsupposable data to model developers and testers---is proposed. UTG uses a variational autoencoder (VAE) to generate unsupposable data. The unsupposable data is generated by acquiring latent values with low occurrence probability in the prior distribution of the VAE and inputting the acquired latent values into the decoder. If unsupposable data is included in the data generated by the decoder, the developer can recognize new unsupposable features by referring to the data. On the basis of those unsupposable features, the developer will be able to create other unsupposable data with the same features. The proposed UTG was applied to the MNIST dataset and the House Sales Price dataset. The results demonstrate the feasibility of UTG.
Open AccessUnsupposable Test-data Generation for Machine-learned Software
As for software development by machine learning, a trained model is evaluated by using part of an existing dataset as test data. However, if data with characteristics that differ from the existing data is input, the model does not always behave as expected. Accordingly, to confirm the behavior of the model more strictly, it is necessary to create data that differs from the existing data and test the model with that different data. The data to be tested includes not only data that developers can suppose (supposable data) but also data they cannot suppose (unsupposable data). To confirm the behavior of the model strictly, it is important to create as much unsupposable data as possible. In this study, therefore, a method called "unsupposable test-data generation" (UTG)---for giving suggestions for unsupposable data to model developers and testers---is proposed. UTG uses a variational autoencoder (VAE) to generate unsupposable data. The unsupposable data is generated by acquiring latent values with low occurrence probability in the prior distribution of the VAE and inputting the acquired latent values into the decoder. If unsupposable data is included in the data generated by the decoder, the developer can recognize new unsupposable features by referring to the data. On the basis of those unsupposable features, the developer will be able to create other unsupposable data with the same features. The proposed UTG was applied to the MNIST dataset and the House Sales Price dataset. The results demonstrate the feasibility of UTG.
- 17 February 2020
- Published by SPIE-Intl Soc Optical Eng
- Vol. 11240, 179
Real-time improvement of LED-based photoacoustic image quality using intermittent pulse echo acquisitions
Photoacoustic imaging is a hybrid modality with advantages of both the ultrasound and optical imaging. This technique with optical spectroscopic contrast and acoustic resolution holds promise in wide range of clinical and preclinical applications. Since photoacoustic imaging also involves ultrasound detection, it is straight forward to implement pointof- care dual mode systems with capability of structural and functional imaging. High signal generated from skin-melanin layer is an important problem in any handheld reflection-mode photoacoustic probes. This problem caused by high light fluence just beneath the ultrasound probe results in reduction of vascular contrast and also causes difficulty in image interpretation. In this work, using our LED-based photoacoustic and ultrasound imaging system (AcousticX), we demonstrate the potential of using ultrasound acquisitions dynamically to suppress high photoacoustic signal from skin surface and qualitatively improve the image contrast and quality. In AcousticX, photoacoustic and conventional pulseecho acquisitions are performed in an intermittent manner resulting in a dual-mode display frame rate of 30 Hz. We make use of the line-by-line/planewave ultrasound acquisitions to automatically delineate the skin surface in quasi realtime and suppress photoacoustic signal from the identified area, resulting in improved image interpretation and contrast without losing temporal resolution. Real-time 2D and 3D (linear scan) imaging experiments were performed on finger and foot dorsum of healthy human volunteers to validate the new skin-signal suppression technique. Results give a direct confirmation that our ultrasound-assisted skin-signal reduction feature holds strong potential in enhancing vascular contrast and visual quality in real-time 2D and 3D LED-based photoacoustic imaging.
- 17 February 2020
- Published by SPIE-Intl Soc Optical Eng
- Vol. 11240, 179
Real-time improvement of LED-based photoacoustic image quality using intermittent pulse echo acquisitions
Photoacoustic imaging is a hybrid modality with advantages of both the ultrasound and optical imaging. This technique with optical spectroscopic contrast and acoustic resolution holds promise in wide range of clinical and preclinical applications. Since photoacoustic imaging also involves ultrasound detection, it is straight forward to implement pointof- care dual mode systems with capability of structural and functional imaging. High signal generated from skin-melanin layer is an important problem in any handheld reflection-mode photoacoustic probes. This problem caused by high light fluence just beneath the ultrasound probe results in reduction of vascular contrast and also causes difficulty in image interpretation. In this work, using our LED-based photoacoustic and ultrasound imaging system (AcousticX), we demonstrate the potential of using ultrasound acquisitions dynamically to suppress high photoacoustic signal from skin surface and qualitatively improve the image contrast and quality. In AcousticX, photoacoustic and conventional pulseecho acquisitions are performed in an intermittent manner resulting in a dual-mode display frame rate of 30 Hz. We make use of the line-by-line/planewave ultrasound acquisitions to automatically delineate the skin surface in quasi realtime and suppress photoacoustic signal from the identified area, resulting in improved image interpretation and contrast without losing temporal resolution. Real-time 2D and 3D (linear scan) imaging experiments were performed on finger and foot dorsum of healthy human volunteers to validate the new skin-signal suppression technique. Results give a direct confirmation that our ultrasound-assisted skin-signal reduction feature holds strong potential in enhancing vascular contrast and visual quality in real-time 2D and 3D LED-based photoacoustic imaging.