Naoto Takeda
Research Subjects
Naoto Takeda has a total of 20 co-authors. Here are the top co-authors by number of shared publications:
- Sonia Chernova
Georgia Institute of Technology, Atlanta, USA
3 shared publications - Yasutaka Nishimura
KDDI Research, Inc.,AI Division,Fujimino,Japan
3 shared publications - Kazushi Ikeda
Division of Information Science, Nara Institute of Science and Technology, Nara, Japan
2 shared publications - Roberto Legaspi
KDDI Research, Inc., Fujimino, Japan
2 shared publications Thomas Plötz
School of Interactive Computing, College of Computing, Georgia Institute of Technology,Atlanta,USA
2 shared publications- Akihiro Kobayashi
KDDI Research, Fujimino-shi, Japan
1 shared publication
Latest Publications
- 15 December 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 158-165
ALLFA: Active Learning Through Label and Feature Augmentation
We introduce a novel feature-based active learning approach that leverages leading feature selection methods from explainable AI (XAI) to elicit user corrections on both the classification label and the leading features contributing to the classification decision. Uniquely, our approach can handle such user corrections, which identify missing and superfluous features in a model-agnostic manner that can be applied to any supervised classification model. Our approach generates augmented samples in a fine-grained manner according to user corrections, and then adds them to the training data to teach relevant and irrelevant features. Our results show that our approach outperforms traditional active learning and a leading baseline in two domains, which includes improvement in classification performance from 0.572 to 0.637 given the same number of queries.
- 15 December 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 158-165
ALLFA: Active Learning Through Label and Feature Augmentation
We introduce a novel feature-based active learning approach that leverages leading feature selection methods from explainable AI (XAI) to elicit user corrections on both the classification label and the leading features contributing to the classification decision. Uniquely, our approach can handle such user corrections, which identify missing and superfluous features in a model-agnostic manner that can be applied to any supervised classification model. Our approach generates augmented samples in a fine-grained manner according to user corrections, and then adds them to the training data to teach relevant and irrelevant features. Our results show that our approach outperforms traditional active learning and a leading baseline in two domains, which includes improvement in classification performance from 0.572 to 0.637 given the same number of queries.
- Open Access
- 13 November 2023
- Published by Association for Computing Machinery (ACM)
- p. 7-10
Modeling and generating human mobility trajectories using transformer with day encoding
Modeling and predicting human mobility trajectories in urban areas is a crucial challenge with various applications. The HuMob Challenge is a competition that focuses on modeling human mobility trajectories using large open-source datasets. This paper presents one of the top 10 winning methods from the competition. In recent years, a method has been developed to apply advanced large language model (LLM) technology to model human movement. We have enhanced trajectory prediction accuracy by encoding daily characteristics, such as commuting to the office on weekdays and visiting shopping districts on weekends, to capture day-to-day variations in movement patterns.
- 13 November 2023
- Published by Association for Computing Machinery (ACM)
- p. 7-10
Open AccessModeling and generating human mobility trajectories using transformer with day encoding
Modeling and predicting human mobility trajectories in urban areas is a crucial challenge with various applications. The HuMob Challenge is a competition that focuses on modeling human mobility trajectories using large open-source datasets. This paper presents one of the top 10 winning methods from the competition. In recent years, a method has been developed to apply advanced large language model (LLM) technology to model human movement. We have enhanced trajectory prediction accuracy by encoding daily characteristics, such as commuting to the office on weekdays and visiting shopping districts on weekends, to capture day-to-day variations in movement patterns.
- Open Access
- 5 May 2023
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Interactive Intelligent Systems
- Vol. 13 (2) , 1-39
Explainable Activity Recognition for Smart Home Systems
Smart home environments are designed to provide services that help improve the quality of life for the occupant via a variety of sensors and actuators installed throughout the space. Many automated actions taken by a smart home are governed by the output of an underlying activity recognition system. However, activity recognition systems may not be perfectly accurate, and therefore inconsistencies in smart home operations can lead users reliant on smart home predictions to wonder “Why did the smart home do that?” In this work, we build on insights from Explainable Artificial Intelligence (XAI) techniques and introduce an explainable activity recognition framework in which we leverage leading XAI methods (Local Interpretable Model-agnostic Explanations, SHapley Additive exPlanations (SHAP), Anchors) to generate natural language explanations that explain what about an activity led to the given classification. We evaluate our framework in the context of a commonly targeted smart home scenario: autonomous remote caregiver monitoring for individuals who are living alone or need assistance. Within the context of remote caregiver monitoring, we perform a two-step evaluation: (a) utilize Machine Learning experts to assess the sensibility of explanations and (b) recruit non-experts in two user remote caregiver monitoring scenarios, synchronous and asynchronous, to assess the effectiveness of explanations generated via our framework. Our results show that the XAI approach, SHAP, has a 92% success rate in generating sensible explanations. Moreover, in 83% of sampled scenarios users preferred natural language explanations over a simple activity label, underscoring the need for explainable activity recognition systems. Finally, we show that explanations generated by some XAI methods can lead users to lose confidence in the accuracy of the underlying activity recognition model, while others lead users to gain confidence. Taking all studied factors into consideration, we make a recommendation regarding which existing XAI method leads to the best performance in the domain of smart home automation and discuss a range of topics for future work to further improve explainable activity recognition.
- 5 May 2023
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Interactive Intelligent Systems
- Vol. 13 (2) , 1-39
Open AccessExplainable Activity Recognition for Smart Home Systems
Smart home environments are designed to provide services that help improve the quality of life for the occupant via a variety of sensors and actuators installed throughout the space. Many automated actions taken by a smart home are governed by the output of an underlying activity recognition system. However, activity recognition systems may not be perfectly accurate, and therefore inconsistencies in smart home operations can lead users reliant on smart home predictions to wonder “Why did the smart home do that?” In this work, we build on insights from Explainable Artificial Intelligence (XAI) techniques and introduce an explainable activity recognition framework in which we leverage leading XAI methods (Local Interpretable Model-agnostic Explanations, SHapley Additive exPlanations (SHAP), Anchors) to generate natural language explanations that explain what about an activity led to the given classification. We evaluate our framework in the context of a commonly targeted smart home scenario: autonomous remote caregiver monitoring for individuals who are living alone or need assistance. Within the context of remote caregiver monitoring, we perform a two-step evaluation: (a) utilize Machine Learning experts to assess the sensibility of explanations and (b) recruit non-experts in two user remote caregiver monitoring scenarios, synchronous and asynchronous, to assess the effectiveness of explanations generated via our framework. Our results show that the XAI approach, SHAP, has a 92% success rate in generating sensible explanations. Moreover, in 83% of sampled scenarios users preferred natural language explanations over a simple activity label, underscoring the need for explainable activity recognition systems. Finally, we show that explanations generated by some XAI methods can lead users to lose confidence in the accuracy of the underlying activity recognition model, while others lead users to gain confidence. Taking all studied factors into consideration, we make a recommendation regarding which existing XAI method leads to the best performance in the domain of smart home automation and discuss a range of topics for future work to further improve explainable activity recognition.
- 13 March 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 659-664
Extraction of Important Temporal Order for eXplainable AI on Time-series data
We propose a new eXplainable AI method on time-series data in which multiple events are arranged in temporal order, to extract important temporal order between events for opaque model decision. Toward this, our proposed method analyzes the model behavior when inputting perturbated data generated by changing the order of the events. We evaluate our method via an exemplary classification task on a simulated dataset to confirm how accurate our method is in extracting important temporal order of events. In addition, we evaluate our method on a smart home sensor dataset to demonstrate what kind of important order is actually extracted.
- 13 March 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 659-664
Extraction of Important Temporal Order for eXplainable AI on Time-series data
We propose a new eXplainable AI method on time-series data in which multiple events are arranged in temporal order, to extract important temporal order between events for opaque model decision. Toward this, our proposed method analyzes the model behavior when inputting perturbated data generated by changing the order of the events. We evaluate our method via an exemplary classification task on a simulated dataset to confirm how accurate our method is in extracting important temporal order of events. In addition, we evaluate our method on a smart home sensor dataset to demonstrate what kind of important order is actually extracted.
- 1 January 2022
- Published by Springer Nature
- p. 3-23
Event Detection and Event-Relevant Tweet Extraction with Human Mobility
- 1 January 2022
- Published by Springer Nature
- p. 3-23
Event Detection and Event-Relevant Tweet Extraction with Human Mobility
- 1 April 2020
- Published by Institute of Electrical Engineers of Japan (IEE Japan) in IEEJ Transactions on Electronics, Information and Systems
- Vol. 140 (4) , 476-483
Detecting Tampered Region in video using LSTM and U-Net
Recently, we can always shoot digital videos with smartphone and share them with family or friends after editing them. It is necessary to check whether they are tampered by someone or not, if we use these digital videos as forensic evidence in criminal investigation. This paper proposes a method for detecting tampered regions in a video where a specific subject is removed from a scene. In the proposed method, we use LSTM for the time domain analysis and U-Net for the space domain analysis. The detection accuracy achieved 0.98 in terms of F-measure, even though tampered region was deformed or moved in the video. The experimental results show superior performance to detect tampered regions for digital videos.
- 1 April 2020
- Published by Institute of Electrical Engineers of Japan (IEE Japan) in IEEJ Transactions on Electronics, Information and Systems
- Vol. 140 (4) , 476-483
Detecting Tampered Region in video using LSTM and U-Net
Recently, we can always shoot digital videos with smartphone and share them with family or friends after editing them. It is necessary to check whether they are tampered by someone or not, if we use these digital videos as forensic evidence in criminal investigation. This paper proposes a method for detecting tampered regions in a video where a specific subject is removed from a scene. In the proposed method, we use LSTM for the time domain analysis and U-Net for the space domain analysis. The detection accuracy achieved 0.98 in terms of F-measure, even though tampered region was deformed or moved in the video. The experimental results show superior performance to detect tampered regions for digital videos.