Yoshihiro Yamada

Tommi Kerola, Yuya Masuda, Takashi Masuko et al.

We introduce PLaMo 2.1-VL, a lightweight Vision Language Model (VLM) for autonomous devices, available in 8B and 2B variants and designed for local and edge deployment with Japanese-language operation. Focusing on Visual Question Answering (VQA) and Visual Grounding as its core capabilities, we develop and evaluate the models for two real-world application scenarios: factory task analysis via tool recognition, and infrastructure anomaly detection. We also develop a large-scale synthetic data generation pipeline and comprehensive Japanese training and evaluation resources. PLaMo 2.1-VL outperforms comparable open models on Japanese and English benchmarks, achieving 61.5 ROUGE-L on JA-VG-VQA-500 and 85.2% accuracy on Japanese Ref-L4. For the two application scenarios, it achieves 53.9% zero-shot accuracy on factory task analysis, and fine-tuning on power plant data improves anomaly detection bbox + label F1-score from 39.7 to 64.9.

Yoshihiro Yamada

Transformers have driven remarkable breakthroughs in natural language processing and computer vision, yet their standard attention mechanism still imposes O(N^2) complexity, hindering scalability to longer sequences. We introduce Circular-convolutional ATtention (CAT), a Fourier-based approach that efficiently applies circular convolutions to reduce complexity without sacrificing representational power. CAT achieves O(NlogN) computations, requires fewer learnable parameters by streamlining fully-connected layers, and introduces no heavier operations, resulting in consistent accuracy improvements and about a 10% speedup in naive PyTorch implementations on large-scale benchmarks such as ImageNet-1k and WikiText-103. Grounded in an engineering-isomorphism framework, CAT's design not only offers practical efficiency and ease of implementation but also provides insights to guide the development of next-generation, high-performance Transformer architectures. Finally, our ablation studies highlight the key conditions underlying CAT's success, shedding light on broader principles for scalable attention mechanisms.

Taiga Kashima, Yoshihiro Yamada, Shunta Saito

The common pipeline of training deep neural networks consists of several building blocks such as data augmentation and network architecture selection. AutoML is a research field that aims at automatically designing those parts, but most methods explore each part independently because it is more challenging to simultaneously search all the parts. In this paper, we propose a joint optimization method for data augmentation policies and network architectures to bring more automation to the design of training pipeline. The core idea of our approach is to make the whole part differentiable. The proposed method combines differentiable methods for augmentation policy search and network architecture search to jointly optimize them in the end-to-end manner. The experimental results show our method achieves competitive or superior performance to the independently searched results.

Md. Rabiul Islam, Shuji Sakamoto, Yoshihiro Yamada et al.

Reading analysis can give important information about a user's confidence and habits and can be used to construct feedback to improve a user's reading behavior. A lack of labeled data inhibits the effective application of fully-supervised Deep Learning (DL) for automatic reading analysis. In this paper, we propose a self-supervised DL method for reading analysis and evaluate it on two classification tasks. We first evaluate the proposed self-supervised DL method on a four-class classification task on reading detection using electrooculography (EOG) glasses datasets, followed by an evaluation of a two-class classification task of confidence estimation on answers of multiple-choice questions (MCQs) using eye-tracking datasets. Fully-supervised DL and support vector machines (SVMs) are used to compare the performance of the proposed self-supervised DL method. The results show that the proposed self-supervised DL method is superior to the fully-supervised DL and SVM for both tasks, especially when training data is scarce. This result indicates that the proposed self-supervised DL method is the superior choice for reading analysis tasks. The results of this study are important for informing the design and implementation of automatic reading analysis platforms.

Yoshihiro Yamada, Masakazu Iwamura, Takuya Akiba et al.

Overfitting is a crucial problem in deep neural networks, even in the latest network architectures. In this paper, to relieve the overfitting effect of ResNet and its improvements (i.e., Wide ResNet, PyramidNet, and ResNeXt), we propose a new regularization method called ShakeDrop regularization. ShakeDrop is inspired by Shake-Shake, which is an effective regularization method, but can be applied to ResNeXt only. ShakeDrop is more effective than Shake-Shake and can be applied not only to ResNeXt but also ResNet, Wide ResNet, and PyramidNet. An important key is to achieve stability of training. Because effective regularization often causes unstable training, we introduce a training stabilizer, which is an unusual use of an existing regularizer. Through experiments under various conditions, we demonstrate the conditions under which ShakeDrop works well.

Yoshihiro Yamada, Masakazu Iwamura, Koichi Kise

On general object recognition, Deep Convolutional Neural Networks (DCNNs) achieve high accuracy. In particular, ResNet and its improvements have broken the lowest error rate records. In this paper, we propose a method to successfully combine two ResNet improvements, ResDrop and PyramidNet. We confirmed that the proposed network outperformed the conventional methods; on CIFAR-100, the proposed network achieved an error rate of 16.18% in contrast to PiramidNet achieving that of 18.29% and ResNeXt 17.31%.