KT
K. Tanahashi
info
Please Note
<p>This page displays the records of the person named above and is not linked to a unique person identifier. This record may need to be merged to a profile.</p>
2 records found
1
Contrastive Self-Supervised Learning for Utility-Privacy Trade-off Navigation in Graph Neural Networks
Node-Level Vulnerability Analysis and DP Utility Recovery
Graphs are everywhere in the real world. Although Graph Neural Networks (GNNs) have demonstrated strong performance across a wide range of graph-based tasks, their unique neighbourhood aggregation mechanism encodes sensitive feature, label, and topological information into model weights, making them vulnerable to Membership Inference Attacks (MIA). This poses challenges for organizations regarding the practical deployment of GNN models trained on their private data, as they have the obligation to protect the privacy of individuals in the dataset. While numerous privacy-preserving techniques, including differential privacy (DP) based approaches such as PrivGNN, have been proposed, they are subject to the inherent utility-privacy trade-off where achieving strong privacy often comes at the cost of reduction in model utility. In this work, we investigate Self-Supervised Learning (SSL) pretraining on a public graph, with a focus on contrastive pretraining via GRACE, as a utility-preserving privacy mitigation strategy with the potential to improve privacy without sacrificing generalizability. All experiments are performed under an inductive setting with disjoint graphs, where the model is evaluated on nodes never seen during training. We first show that SSL improves the utility-privacy trade-off in standard GNN training, with the backbone learning rate (BLR) acting as a tunable knob to navigate the trade-off. Subsequently, we perform a node-level vulnerability analysis, finding that structural isolation and intra-class feature dissimilarity act as predictors for nodes' privacy vulnerability, while neighbourhood class divergence exhibits more dataset-dependent results. We then show that employing lower BLR alone can disproportionately reduce the privacy vulnerability of the most exposed nodes, with SSL providing additional, uniform privacy benefit on top of it. Beyond standard GNN training, we integrate SSL into DP-based PrivGNN and show that it substantially recovers the utility lost to privacy mechanisms, providing significantly tighter formal privacy guarantees without hurting empirical privacy.
...
Graphs are everywhere in the real world. Although Graph Neural Networks (GNNs) have demonstrated strong performance across a wide range of graph-based tasks, their unique neighbourhood aggregation mechanism encodes sensitive feature, label, and topological information into model weights, making them vulnerable to Membership Inference Attacks (MIA). This poses challenges for organizations regarding the practical deployment of GNN models trained on their private data, as they have the obligation to protect the privacy of individuals in the dataset. While numerous privacy-preserving techniques, including differential privacy (DP) based approaches such as PrivGNN, have been proposed, they are subject to the inherent utility-privacy trade-off where achieving strong privacy often comes at the cost of reduction in model utility. In this work, we investigate Self-Supervised Learning (SSL) pretraining on a public graph, with a focus on contrastive pretraining via GRACE, as a utility-preserving privacy mitigation strategy with the potential to improve privacy without sacrificing generalizability. All experiments are performed under an inductive setting with disjoint graphs, where the model is evaluated on nodes never seen during training. We first show that SSL improves the utility-privacy trade-off in standard GNN training, with the backbone learning rate (BLR) acting as a tunable knob to navigate the trade-off. Subsequently, we perform a node-level vulnerability analysis, finding that structural isolation and intra-class feature dissimilarity act as predictors for nodes' privacy vulnerability, while neighbourhood class divergence exhibits more dataset-dependent results. We then show that employing lower BLR alone can disproportionately reduce the privacy vulnerability of the most exposed nodes, with SSL providing additional, uniform privacy benefit on top of it. Beyond standard GNN training, we integrate SSL into DP-based PrivGNN and show that it substantially recovers the utility lost to privacy mechanisms, providing significantly tighter formal privacy guarantees without hurting empirical privacy.
Agent Failure, Trust Repair, and Fluency in Human-AI Teams
Impact of Opportunistic Interdependence Relationship on Trust Violation, Trust Repair, and on Collaboration Fluency in a Human-Agent Team
Nowadays, Human Autonomy Teams (HATs) are incorporated in many fields, where humans and autonomous agents work collaboratively to combine their capabilities with the ultimate goal of performing tasks more efficiently. In such environments, it is imperative to sustain a high level of trust between the agents as collaboration is not possible without mutual trust. Naturally, this implies that recovering trust following trust violation is also a crucial aspect of HATs. Moreover, besides team performance, the fluency of collaboration is another important factor to consider when evaluating the success of the teams. This paper aims to investigate the effect of opportunistic (soft) interdependence between the agents on trust violation, trust repair, and on collaboration fluency when compared against a baseline (complete independence) condition. In this paper, interdependence relationships refer to how the agents complement/combine each other's competence. The experimental results were obtained through a user study, using questionnaires and logged objective metrics. Our research found that teams with opportunistic interdependence relationships were significantly affected by trust violations compared to the baseline condition. Furthermore, although not as significant as the effect of trust violation, they also experienced a significant trust recovery during the tasks. Finally, the results of analyzing both subjective and objective fluency metrics did not give any significant result that indicates the difference in the level of collaboration fluency between the two conditions.
...
Nowadays, Human Autonomy Teams (HATs) are incorporated in many fields, where humans and autonomous agents work collaboratively to combine their capabilities with the ultimate goal of performing tasks more efficiently. In such environments, it is imperative to sustain a high level of trust between the agents as collaboration is not possible without mutual trust. Naturally, this implies that recovering trust following trust violation is also a crucial aspect of HATs. Moreover, besides team performance, the fluency of collaboration is another important factor to consider when evaluating the success of the teams. This paper aims to investigate the effect of opportunistic (soft) interdependence between the agents on trust violation, trust repair, and on collaboration fluency when compared against a baseline (complete independence) condition. In this paper, interdependence relationships refer to how the agents complement/combine each other's competence. The experimental results were obtained through a user study, using questionnaires and logged objective metrics. Our research found that teams with opportunistic interdependence relationships were significantly affected by trust violations compared to the baseline condition. Furthermore, although not as significant as the effect of trust violation, they also experienced a significant trust recovery during the tasks. Finally, the results of analyzing both subjective and objective fluency metrics did not give any significant result that indicates the difference in the level of collaboration fluency between the two conditions.