Search

Recently some studies have highlighted the potential of Large Language Models (LLMs) as effective generators of supervised training data, offering advantages such as enhanced inference efficiency and reduced costs associated with data collection. However, these studies have predominantly focused on English language tasks. In this paper, we address the fundamental research question: Can LLMs serve as proficient training data generators for other language tasks? Specifically, we leverage LLMs to synthesize supervised training data under few-shot and zero-shot learning scenarios across six diverse Japanese downstream tasks. Subsequently, we utilize this synthesized data to train compact models (e.g., BERT). This novel methodology is termed JAPAGEN. Our experimental findings underscore that JAPAGEN achieves robust performance in classification tasks that necessitate formal text inputs, demonstrating competitive results compared to conventional LLM prompting strategies., Comment: Accepted by PACLIC38 (2024)

Infertility is a global health problem, and an increasing number of couples are seeking medical assistance to achieve reproduction, at least half of which are caused by men. The success rate of assisted reproductive technologies depends on sperm assessment, in which experts determine whether sperm can be used for reproduction based on morphology and motility of sperm. Previous sperm assessment studies with deep learning have used datasets comprising images that include only sperm heads, which cannot consider motility and other morphologies of sperm. Furthermore, the labels of the dataset are one-hot, which provides insufficient support for experts, because assessment results are inconsistent between experts, and they have no absolute answer. Therefore, we constructed the video dataset for sperm assessment whose videos include sperm head as well as neck and tail, and its labels were annotated with soft-label. Furthermore, we proposed the sperm assessment framework and the neural network, RoSTFine, for sperm video recognition. Experimental results showed that RoSTFine could improve the sperm assessment performances compared to existing video recognition models and focus strongly on important sperm parts (i.e., head and neck)., Comment: Accepted at Winter Conference on Applications of Computer Vision (WACV) 2024

Text-based person re-identification (TBPReID) aims to retrieve person images represented by a given textual query. In this task, how to effectively align images and texts globally and locally is a crucial challenge. Recent works have obtained high performances by solving Masked Language Modeling (MLM) to align image/text parts. However, they only performed uni-directional (i.e., from image to text) local-matching, leaving room for improvement by introducing opposite-directional (i.e., from text to image) local-matching. In this work, we introduce Bidirectional Local-Matching (BiLMa) framework that jointly optimize MLM and Masked Image Modeling (MIM) in TBPReID model training. With this framework, our model is trained so as the labels of randomly masked both image and text tokens are predicted by unmasked tokens. In addition, to narrow the semantic gap between image and text in MIM, we propose Semantic MIM (SemMIM), in which the labels of masked image tokens are automatically given by a state-of-the-art human parser. Experimental results demonstrate that our BiLMa framework with SemMIM achieves state-of-the-art Rank@1 and mAP scores on three benchmarks., Comment: Accepted at ICCVW 2023

This paper investigates the effect of tokenizers on the downstream performance of pretrained language models (PLMs) in scriptio continua languages where no explicit spaces exist between words, using Japanese as a case study. The tokenizer for such languages often consists of a morphological analyzer and a subword tokenizer, requiring us to conduct a comprehensive study of all possible pairs. However, previous studies lack this comprehensiveness. We therefore train extensive sets of tokenizers, build a PLM using each, and measure the downstream performance on a wide range of tasks. Our results demonstrate that each downstream task has a different optimal morphological analyzer, and that it is better to use Byte-Pair-Encoding or Unigram rather than WordPiece as a subword tokenizer, regardless of the type of task., Comment: Accepted at ACL SRW 2023

Controlling gain and loss of coupled optical cavities can induce non-Hermitian degeneracies of eigenstates, called exceptional points (EPs). Various unconventional phenomena around EPs have been reported, and expected to incorporate extra functionalities into photonic devices. The eigenmode exactly under the EP degeneracy is also predicted to exhibit enhanced radiation. However, such responses have yet to be observed in on-chip lasers, because of both the limited controllability of their gain and loss and the lifting of degeneracy by pump-induced cavity detuning. Here, we report the first non-Hermitian nanophotonic platform based on two electrically pumped photonic crystal lasers and its spontaneous emission at an EP degeneracy. Systematically tuned and independent current injection to our wavelength-scale active heterostructure cavities enables us to demonstrate the clear EP phase transition of their spontaneous emission, accompanied with the spectral coalescence of coupled modes and reversed pump dependence of the intensity. Furthermore, we find experimentally and confirm theoretically the peculiar squared Lorentzian emission spectrum very near the exact EP, which indicates the four-fold enhancement of the photonic local density of states induced purely by the degeneracy. Our results open a new pathway to engineer the light-matter interaction by non-Hermiticity and explore larger reconfigurable laser arrays for further non-Hermitian features and physics., Comment: 13 pages, 5 figures, and supplementary materials