| 谷内 洋介 | D, 中間発表 | 数理情報学 | 池田 和司 | 松本 健一 | 久保 孝富 | 日永田 智絵 | Li Yuzhe | |
|
title: A Study on Social Comparison Mechanisms: A Probabilistic Generative Modeling Approach
abstract: Social comparison refers to the cognitive process of comparing one's own rewards with those of others, a common element in definitions of social emotions such as envy. Recent neuroscientific studies on monkeys have revealed that even when the probability of receiving one's own reward remains constant, the subjective value of the reward "for oneself" decreases as the probability of the partner receiving a reward increases. This study investigated whether monkeys estimate the subjective value of rewards "for their partner" during social comparison. Given recent monkey studies suggesting the presence of theory of mind, it was hypothesized that monkeys might also estimate the subjective value for their partners. For verification, we trained mMLDA, an interpretable multimodal probabilistic generative model, on monkey data. We compared three models - one that estimates the partner's subjective value and one that does not and the other control model - to determine which better reproduces the fluctuations in one's own subjective value caused by the partner's rewards. The results showed that the model that did not consider the partner's subjective value performed better at reproducing the observed patterns. This suggests that monkeys do not take into account their partner's subjective value during social comparison. language of the presentation: Japanese 発表題目: 確率的生成モデルによる社会比較メカニズムの調査 発表概要: 社会比較とは、自分と他者の報酬を比較する認知プロセスを指し、嫉妬などの社会的情動の定義に共通して見られる。近年のサルに対する神経科学的研究では、自分に報酬が与えられる確率が一定でも、他者が報酬を得る確率が上がるにつれて、報酬の「自分にとっての」主観的価値が下がることが判明している。本研究では、社会比較において、サルが「相手にとっての」報酬の主観的価値を推測しているかどうかを計算論的アプローチによって検証した。近年のサル研究では、サルが心の理論を持つことが示唆されていることから、他者にとっての主観的価値も推測されていると期待される。検証においては、解釈性の高いマルチモーダルな確率的生成モデルである、mMLDAを用いてサルのデータを学習した。他者の主観的価値を推測するモデルとそうでないモデルで、どちらが他者の報酬による自分の主観的価値の変動を上手く再現できるかを検討したところ、他者の主観的価値を考慮に入れないモデルの方が上手く再現できていた。サルが社会比較において相手の主観的価値を考慮に入れていないことが示唆された。 | ||||||||
| 中川 郁仁 | M, 1回目発表 | 数理情報学 | 池田 和司 | 田中 沙織 | 久保 孝富 | 日永田 智絵 | Li Yuzhe | |
|
title: Investigation of Emotional Responses to Gustatory and Visual Stimuli
abstract:Emotion is known to be a vital function involved in human decision-making and various studies have investigated emotions elicited by visual and auditory stimuli. However, emotions often arise from stimuli across multiple modalities, and there is limited research on emotions triggered by multimodal stimuli. This study focuses on eating behavior, which involves stimuli from multiple modalities, and examines emotions elicited by gustatory stimuli as well as those influenced by the combination of gustatory stimuli and visual stimuli (color). language of the presentation:Japanese 発表題目:味覚と視覚の刺激による情動反応の調査 発表概要:情動は人の意思決定に関わっているなど、生きるために必要な機能であることが知られており、視覚や聴覚の刺激による情動について様々な研究で調べられている。しかし、情動は複数モダリティからの刺激から現れることが多く複数モダリティから得られる刺激による情動については調べられている研究は少ない。本研究では、複数モダリティの刺激が得られる食行動に着目し、味覚刺激による情動と味覚刺激と視覚刺激(色)による情動を調べる。 | ||||||||
| 廣中 高太郎 | M, 1回目発表 | 数理情報学 | 池田 和司 | 田中 沙織 | 久保 孝富 | 日永田 智絵 | Li Yuzhe | |
|
title: Analysis of local field potentials in the dorsal premotor cortex of the monkey during a shape-manipulation task abstract: Analyses of the spiking activity of neurons in the dorsal premotor cortex (PMd) suggest that the PMd plays an important role in action selection based on cognitive information. However, how local field potentials (LFPs) are modulated during action selection has not been extensively investigated. Here, we examined task-dependent changes in LFPs recorded from the PMd while monkeys performed a shape manipulation task that required them to plan stepwise actions based on perceptual information about visual shape. Language of the presentation: English 発表題目: 形操作課題遂行中のサル背側運動前野の局所場電位の解析 発表概要:背足運動前野（PMd）におけるニューロンのスパイク活動の解析から、PMdが認知情報に基づいた行動選択において重要な役割を果たしていることが示唆される。しかし、行動選択の過程で局所電場電位（LFP）がどのように変調されるかについては、これまで十分に調査されていない。本研究では、視覚的な形状情報に基づいて段階的な動作を計画する必要がある形状操作課題をサルが実施する際に、PMdから記録したLFPの課題依存的な変化を調べた。 | ||||||||
| 本間 天譲 | M, 2回目発表 | ロボットラーニング | 松原 崇充 | 安本 慶一 | 柴田 一騎 | 鶴峯 義久 | 佐々木 光 | 角川 勇貴 |
|
title: Sim-to-Real Reinforcement Learning for Complex Tasks on Neurochip-Driven Edge Robots
abstract: Neurochips are a computing device suitable for function approximation of spiking neural networks (SNNs), and because of their energy-saving and real-time performance, they have attracted attention as a computing device for control policies in autonomous control tasks of edge robots. Sim-to-real reinforcement learning is useful as a method for acquiring control policies for edge robots, but because SNNs approximate functions with a set of spike signals, learning with domain randomization is difficult due to the accuracy of the approximation. In previous studies on robot control policy acquisition using SNNs, methods have been developed to learn SNNs using gap amplification and distillation, but the instability of learning in domain randomization remains an issue, and Sim-to-Real has not been achieved. In this study, therefore, to solve this issue, we propose an SNN policy update method that combines gap amplification and distillation to strengthen the approximation ability of SNNs and improve the stability and accuracy of policy updates. The contributions of this study are to realize Sim-to-Real using SNNs for the first time, propose a new learning method that combines gap amplification and distillation, and demonstrate its effectiveness. As an experiment, we verified the effectiveness of this method in a maze task, and the results showed that the proposed method outperformed conventional methods, confirming its effectiveness in sim-to-real tasks. language of the presentation: Japanese | ||||||||
| 知念 朋輝 | M, 1回目発表 | 大規模システム管理 | 笠原 正治 | 林 優一 | 原 崇徳 | 中畑 裕 |
|
title: Blockchain-based Anonymous Two-Factor Biometric Authentication for IoT Devices
abstract: With the widespread adoption of IoT devices, the need to protect their confidentiality, integrity, and availability from potential attackers has become increasingly critical. However, security challenges, primarily caused by unauthorized access, remain unresolved. This study proposes a blockchain-based two-factor biometric authentication system designed to protect users' identities and biometric data, while also providing resilience against common authentication attacks. The proposed approach combines zero-knowledge membership proofs with zkSNARK to ensure that biometric data is not exposed to third parties. Furthermore, the approach investigates an efficient design suitable for high-traffic IoT environments, aiming to optimize the trade-off between security, privacy, and throughput in the development of a next-generation authentication framework. language of the presentation: Japanese | ||||||
| 井上 明浩 | M, 1回目発表 | 大規模システム管理 | 笠原 正治 | 藤川 和利 | 原 崇徳 | |
|
title: Training Latency Aware Resource Allocation Problem in Parallel Split Learning
abstract: Federated learning facilitates the training of machine learning (ML) models in a decentralized nature of data, enhancing data privacy. However, the participation of resource-limited edge devices (e.g., IoT devices) causes delays in the training process. ML models with a huge number of parameters require computing, memory, and communication resources that exceed the capabilities of edge devices. To address these issues, parallel split learning has been proposed, allowing multiple resource-limited nodes to participate in joint training processes with assistance from resourceful computing nodes. To minimize the training latency, we need to solve resource allocation problems (i.e., cut layer selection, communication channel allocation, and power control) in parallel split learning. This presentation introduces a novel approach to the resource allocation problems and clarifies the research challenges. language of the presentation: Japanese | ||||||
| 池田 遼太 | M, 1回目発表 | 大規模システム管理 | 笠原 正治 | 松本 健一 | 原 崇徳 | 中畑 裕 |
|
title: Collusion Resistance Analysis of DAO Voting Mechanisms Based on Bribery Cost Evaluation for Multiple Voting Rounds
abstract: Decentralized Autonomous Organization (DAO) is a novel organizational structure operating on a blockchain, where decision-making is conducted through voting by all participants, distinguishing it from traditional organizations. However, concerns have been raised regarding the loss of decentralization when large token holders dominate the voting process.Existing research has compared the collusion resistance of three voting mechanisms: Linear Voting, Quadratic Voting, and the veToken Mechanism. However, this research assumes unrealistic conditions, such as restricting DAO voting to a single round and excluding large token holders from participation, which fail to accurately reflect the real-world environment surrounding DAO.To address these limitations, this research extends the existing models by incorporating scenarios where large token holders actively participate in voting and where voting occurs over multiple rounds. Using bribery costs as a key metric, the research evaluates the collusion resistance of various voting mechanisms. language of the presentation: Japanese 発表題目: 複数回投票を想定した賄賂コスト評価によるDAO投票メカニズムの談合耐性分析 発表概要: DAO (Decentralized Autonomous Organization) とはブロックチェーン上で運営される新しい組織形態を指し、従来型の組織と異なり参加者全員の投票により意思決定が行われるとして注目を集めている。しかし、大規模トークン保有者が投票を支配することで分散性が損なわれるという課題が指摘されている。既存研究では、Linear Voting、Quadratic Voting、veToken Mechanismの3つの投票メカニズムについて、談合耐性を比較している。しかし、DAOの投票が1回に限定されていることや、大規模トークン保有者は投票には不参加という非現実的な仮定が含まれており、実際のDAOを取り巻く環境を正確に反映しているとは言えない。本研究では、この課題を解決するため、大規模トークン保有者が投票に参加することや投票が複数回行われることを想定したモデルに拡張し、賄賂コストを指標として各投票メカニズムにおける談合耐性を評価する。 | ||||||
| 山田 純也 | M, 1回目発表 | 大規模システム管理 | 笠原 正治 | 松本 健一 | 原 崇徳 | 中畑 裕 |
|
title: Reward Sharing Scheme for Stake Pools in PoS Blockchains Considering Delegators' Cognitive Bias
abstract: In the Proof of Stake (PoS) mechanism currently employed by Cardano, there are stake pool operators responsible for block generation and delegators who delegate their stakes to these pools. This mechanism allows staking with small amounts, increasing the total stake in the network and thereby enhancing its security. While reward sharing schemes utilizing game theory have been proposed to govern the relationship between pool operators and delegators, it is unrealistic to assume that delegators, being human, will always select the optimal pool based purely on theoretical rationality. This study aims to propose a reward sharing scheme that incorporates Prospect Theory from behavioral economics, taking into account human behavioral biases, with the goal of further improving the fairness and decentralization of the network. This presentation will provide an overview of existing research, the rationale for introducing Prospect Theory, and the plans for future research. language of the presentation: Japanese 発表題目: デリゲータの認知バイアスを考慮したPoS型ブロックチェーンにおける報酬共有設計 発表概要: 現在カルダノで使用されているProof of Stake(PoS)では、ブロック生成を担うプール運営者と、プールにステークを委任するデリゲータが存在する。この仕組みにより、少額からステーク可能になり、ネットワーク全体のステーク量が増加することによってネットワークのセキュリティが向上する。ゲーム理論を応用したプール運営者とデリゲータ間の報酬共有方式は提案されているが、実際にはデリゲータは人間であるため、完璧に理論に従って最適なプールを選択するとは言えない。そこで、本研究は報酬共有方式に行動経済学におけるプロスペクト理論を適用することで、人間の行動バイアスを考慮した方式を提案することを目指し、ネットワークの公平性と分散化をさらに向上させることを目的とする。本発表では既存研究の紹介とプロスペクト理論の導入の背景、今後の研究計画について紹介を行う。 | ||||||