| 日時(Datetime) | 令和7年6月17日(火)3限 (13:30--15:00), 2025/06/17, Tueday, 3rd slot |
|---|---|
| 場所(Location) | パナソニックISセミナーホール（L1）, Panasonic IS Seminar Hall （L1） |
| 司会(Chair) | 織田 (Orita) sensei |
| 講演者(Presenter) | Sumitaka Honji, ヒューマン・ロボティクス研究室 (Human Robotics Research Lab.) |
|---|---|
| 題目(Title) | Stochastic modeling and control of soft robots |
| 概要(Abstract) | The flexibility of soft robots brings a lot of advantages, such as physical safety, environmental adaptability, and so on. However, their motion is highly nonlinear and varying, which makes it difficult to model and control the behaviors of soft robots. The stochastic approach is one of the keys to this problem. This talk introduces our methodology of integrating a conventional mechanical model with stochasticity and its expansion to the control objectives. |
| 講演者(Presenter) | Taisho Sasada, ディペンダブルシステム学研究室 (Dependable System Research Lab.) |
|---|---|
| 題目(Title) | A Secure Data Aggregation against Poisoning Attack for Encrypted Data with Differential Privacy |
| 概要(Abstract) | Location data has the potential to reveal congestion patterns during specific times of day and days of the week. By aggregating location data across different organizations, valuable insights can be derived that would be difficult to obtain independently. However, when organizations share data for aggregation, there is a risk of exposing sensitive information. In this research, we introduce a protocol that enables data aggregation across various organizations while maintaining data encryption throughout the entire process. The proposed protocol utilizes Somewhat Homomorphic Encryption to confidentially merge only the records common to two datasets to generate an aggregated table. Subsequently, our approach introduces encrypted noise for differential privacy (DP) to the resulting table, ensuring that DP guarantees are maintained even after decryption. However, applying DP to encrypted data carries the risk of adversaries injecting manipulated data at their discretion. To counter the potential mixing of manipulated and encrypted data, we have developed an algorithm within our proposed protocol to validate the content of encrypted data. |
| 講演者(Presenter) | Araya Kibrom Desta, 情報基盤システム学研究室 (Internet Architecture and Systems Research Lab.) |
|---|---|
| 題目(Title) | Evaluating Transformer Models for Road Traffic Volume Forecasting with Weather-Aware Inputs |
| 概要(Abstract) | Transformer models, originally developed for natural language processing, have recently gained attention in time series forecasting. In this study, we extend their application to road traffic volume prediction as part of efforts to mitigate urban congestion. We utilize five years of sensor data collected from three major traffic junctions in Istanbul, Turkey, and evaluate the performance of several Transformer-based models, including Informer, Autoformer, Reformer, and the recently proposed iTransformer. Our experiments show that the iTransformer consistently outperforms other Transformer variants and also surpasses our baseline models of LSTM and a simple linear models. Additionally, we investigate the effect of incorporating weather information, specifically temperature and categorical weather conditions, on forecasting performance. Across the three junctions, our best-performing models achieved mean absolute errors ranging from 0.107 to 0.284 for short-term forecasts over an 8-step prediction timesteps. |