Hideki Takase

Ai Nozaki, Takuya Kojima, Hideki Takase et al.

Fully Homomorphic Encryption (FHE) enables secure computation over encrypted data, but its computational cost remains a major obstacle to practical deployment. To mitigate this overhead, many studies have explored GPU acceleration for the CKKS scheme, which is widely used for approximate arithmetic. In CKKS, CKKS parameters are configured for each workload by balancing multiplicative depth, security requirements, and performance. These parameters significantly affect ciphertext size, thereby determining how the memory footprint fits within the GPU memory hierarchy. Nevertheless, prior studies typically apply their proposed optimization methods uniformly, without considering differences in CKKS parameter configurations. In this work, we demonstrate that the optimal GPU optimization strategy for CKKS depends on the CKKS parameter configuration. We first classify prior optimizations by two aspects of dataflows which affect memory footprint and then conduct both qualitative and quantitative performance analyses. Our analysis shows that even on the same GPU architecture, the optimal strategy varies with CKKS parameters with performance differences of up to 1.98 $\times$ between strategies, and that the criteria for selecting an appropriate strategy differ across GPU architectures.

Yuki Oda, Yuta Ono, Hiroshi Nakamura et al.

The continuous scaling of deep neural networks has fundamentally transformed machine learning, with larger models demonstrating improved performance across diverse tasks. This growth in model size has dramatically increased the computational resources required for the training process. Consequently, distributed approaches, such as Federated Learning and Split Learning, have become essential paradigms for scalable deployment. However, existing Split Learning approaches assume client homogeneity and uniform split points across all participants. This critically limits their applicability to real-world IoT systems where devices exhibit heterogeneity in computational resources. To address this limitation, this paper proposes Hetero-SplitEE, a novel method that enables heterogeneous IoT devices to train a shared deep neural network in parallel collaboratively. By integrating heterogeneous early exits into hierarchical training, our approach allows each client to select distinct split points (cut layers) tailored to its computational capacity. In addition, we propose two cooperative training strategies, the Sequential strategy and the Averaging strategy, to facilitate this collaboration among clients with different split points. The Sequential strategy trains clients sequentially with a shared server model to reduce computational overhead. The Averaging strategy enables parallel client training with periodic cross-layer aggregation. Extensive experiments on CIFAR-10, CIFAR-100, and STL-10 datasets using ResNet-18 demonstrate that our method maintains competitive accuracy while efficiently supporting diverse computational constraints, enabling practical deployment of collaborative deep learning in heterogeneous IoT ecosystems.

Yuta Ono, Hiroshi Nakamura, Hideki Takase

Federated Active Learning (FAL) seeks to reduce the burden of annotation under the realistic constraints of federated learning by leveraging Active Learning (AL). As FAL settings make it more expensive to obtain ground truth labels, FAL strategies that work well in low-budget regimes, where the amount of annotation is very limited, are needed. In this work, we investigate the effectiveness of TypiClust, a successful low-budget AL strategy, in low-budget FAL settings. Our empirical results show that TypiClust works well even in low-budget FAL settings contrasted with relatively low performances of other methods, although these settings present additional challenges, such as data heterogeneity, compared to AL. In addition, we show that FAL settings cause distribution shifts in terms of typicality, but TypiClust is not very vulnerable to the shifts. We also analyze the sensitivity of TypiClust to feature extraction methods, and it suggests a way to perform FAL even in limited data situations.