Search Results for author: Hideki Nakayama

Found 72 papers, 17 papers with code

Faster AutoAugment: Learning Augmentation Strategies using Backpropagation

1 code implementation ECCV 2020 Ryuichiro Hataya, Jan Zdenek, Kazuki Yoshizoe, Hideki Nakayama

In this paper, we propose a differentiable policy search pipeline for data augmentation, which is much faster than previous methods.

Data Augmentation

Latent-Variable Non-Autoregressive Neural Machine Translation with Deterministic Inference Using a Delta Posterior

1 code implementation20 Aug 2019 Raphael Shu, Jason Lee, Hideki Nakayama, Kyunghyun Cho

By decoding multiple initial latent variables in parallel and rescore using a teacher model, the proposed model further brings the gap down to 1. 0 BLEU point on WMT'14 En-De task with 6. 8x speedup.

Machine Translation Translation

Semantic Aware Attention Based Deep Object Co-segmentation

3 code implementations16 Oct 2018 Hong Chen, Yifei HUANG, Hideki Nakayama

Object co-segmentation is the task of segmenting the same objects from multiple images.

Object Segmentation

OSSGAN: Open-Set Semi-Supervised Image Generation

1 code implementation CVPR 2022 Kai Katsumata, Duc Minh Vo, Hideki Nakayama

We introduce a challenging training scheme of conditional GANs, called open-set semi-supervised image generation, where the training dataset consists of two parts: (i) labeled data and (ii) unlabeled data with samples belonging to one of the labeled data classes, namely, a closed-set, and samples not belonging to any of the labeled data classes, namely, an open-set.

Image Generation

Word Ordering as Unsupervised Learning Towards Syntactically Plausible Word Representations

1 code implementation IJCNLP 2017 Noriki Nishida, Hideki Nakayama

The research question we explore in this study is how to obtain syntactically plausible word representations without using human annotations.

Dependency Parsing Part-Of-Speech Tagging +1

Annotation Order Matters: Recurrent Image Annotator for Arbitrary Length Image Tagging

1 code implementation18 Apr 2016 Jiren Jin, Hideki Nakayama

In addition to comparing our model with existing methods using the conventional top-k evaluation measures, we also provide our model as a high quality baseline for the arbitrary length image tagging task.

Image Captioning Machine Translation +4

StoryER: Automatic Story Evaluation via Ranking, Rating and Reasoning

1 code implementation16 Oct 2022 Hong Chen, Duc Minh Vo, Hiroya Takamura, Yusuke Miyao, Hideki Nakayama

Existing automatic story evaluation methods place a premium on story lexical level coherence, deviating from human preference.

Comment Generation

LED: A Dataset for Life Event Extraction from Dialogs

1 code implementation17 Apr 2023 Yi-Pei Chen, An-Zi Yen, Hen-Hsen Huang, Hideki Nakayama, Hsin-Hsi Chen

Our proposed life event dialog dataset and in-depth analysis of IE frameworks will facilitate future research on life event extraction from conversations.

Event Extraction Relation Extraction

Character-Centric Story Visualization via Visual Planning and Token Alignment

2 code implementations16 Oct 2022 Hong Chen, Rujun Han, Te-Lin Wu, Hideki Nakayama, Nanyun Peng

This task requires machines to 1) understand long text inputs and 2) produce a globally consistent image sequence that illustrates the contents of the story.

Story Visualization Text-to-Image Generation

Partition-and-Debias: Agnostic Biases Mitigation via A Mixture of Biases-Specific Experts

1 code implementation ICCV 2023 Jiaxuan Li, Duc Minh Vo, Hideki Nakayama

However, most of these methods implicitly assume that a given image contains only one type of known or unknown bias, failing to consider the complexities of real-world biases.

Image Classification

Pixel to Binary Embedding Towards Robustness for CNNs

1 code implementation13 Jun 2022 Ikki Kishida, Hideki Nakayama

There are approaches to replace pixel values with binary embeddings to tackle the problem of adversarial perturbations, which successfully improve robustness.

Towards Parameter-Efficient Integration of Pre-Trained Language Models In Temporal Video Grounding

1 code implementation26 Sep 2022 Erica K. Shimomoto, Edison Marrese-Taylor, Hiroya Takamura, Ichiro Kobayashi, Hideki Nakayama, Yusuke Miyao

This paper explores the task of Temporal Video Grounding (TVG) where, given an untrimmed video and a natural language sentence query, the goal is to recognize and determine temporal boundaries of action instances in the video described by the query.

Benchmarking Natural Language Queries +2

Weakly Supervised Formula Learner for Solving Mathematical Problems

1 code implementation COLING 2022 Yuxuan Wu, Hideki Nakayama

Existing work suggested solving this task with a two-phase approach, where the model first predicts formulas from questions and then calculates answers from such formulas.

Mathematical Reasoning Question Answering +1

Deep Learning for Forecasting Stock Returns in the Cross-Section

no code implementations3 Jan 2018 Masaya Abe, Hideki Nakayama

Many studies have been undertaken by using machine learning techniques, including neural networks, to predict stock returns.

BIG-bench Machine Learning speech-recognition +1

Parameter Reference Loss for Unsupervised Domain Adaptation

no code implementations20 Nov 2017 Jiren Jin, Richard G. Calland, Takeru Miyato, Brian K. Vogel, Hideki Nakayama

Unsupervised domain adaptation (UDA) aims to utilize labeled data from a source domain to learn a model that generalizes to a target domain of unlabeled data.

Model Selection Unsupervised Domain Adaptation

Zero-resource Machine Translation by Multimodal Encoder-decoder Network with Multimedia Pivot

no code implementations14 Nov 2016 Hideki Nakayama, Noriki Nishida

We propose an approach to build a neural machine translation system with no supervised resources (i. e., no parallel corpora) using multimodal embedded representation over texts and images.

Machine Translation Translation

Single-Queue Decoding for Neural Machine Translation

1 code implementation6 Jul 2017 Raphael Shu, Hideki Nakayama

Neural machine translation models rely on the beam search algorithm for decoding.

Machine Translation Translation

Later-stage Minimum Bayes-Risk Decoding for Neural Machine Translation

no code implementations11 Apr 2017 Raphael Shu, Hideki Nakayama

For extended periods of time, sequence generation models rely on beam search algorithm to generate output sequence.

Machine Translation Translation

Discrete Structural Planning for Neural Machine Translation

no code implementations14 Aug 2018 Raphael Shu, Hideki Nakayama

Structural planning is important for producing long sentences, which is a missing part in current language generation models.

Machine Translation Sentence +2

Real-time Neural-based Input Method

no code implementations ICLR 2019 Jiali Yao, Raphael Shu, Xinjian Li, Katsutoshi Ohtsuki, Hideki Nakayama

The input method is an essential service on every mobile and desktop devices that provides text suggestions.

Language Modelling

Improving Beam Search by Removing Monotonic Constraint for Neural Machine Translation

no code implementations ACL 2018 Raphael Shu, Hideki Nakayama

However, as the algorithm produces hypotheses in a monotonic left-to-right order, a hypothesis can not be revisited once it is discarded.

Language Modelling Machine Translation +1

Coherence Modeling Improves Implicit Discourse Relation Recognition

no code implementations WS 2018 Noriki Nishida, Hideki Nakayama

The research described in this paper examines how to learn linguistic knowledge associated with discourse relations from unlabeled corpora.

Implicit Discourse Relation Classification Relation +1

Investigating CNNs' Learning Representation under label noise

no code implementations ICLR 2019 Ryuichiro Hataya, Hideki Nakayama

Deep convolutional neural networks (CNNs) are known to be robust against label noise on extensive datasets.

Memorization

Infinite Brain MR Images: PGGAN-based Data Augmentation for Tumor Detection

no code implementations29 Mar 2019 Changhee Han, Leonardo Rundo, Ryosuke Araki, Yujiro Furukawa, Giancarlo Mauri, Hideki Nakayama, Hideaki Hayashi

Due to the lack of available annotated medical images, accurate computer-assisted diagnosis requires intensive Data Augmentation (DA) techniques, such as geometric/intensity transformations of original images; however, those transformed images intrinsically have a similar distribution to the original ones, leading to limited performance improvement.

Data Augmentation

Learning More with Less: GAN-based Medical Image Augmentation

no code implementations29 Mar 2019 Changhee Han, Kohei Murao, Shin'ichi Satoh, Hideki Nakayama

Convolutional Neural Network (CNN)-based accurate prediction typically requires large-scale annotated training data.

Image Augmentation object-detection +1

USE-Net: incorporating Squeeze-and-Excitation blocks into U-Net for prostate zonal segmentation of multi-institutional MRI datasets

no code implementations17 Apr 2019 Leonardo Rundo, Changhee Han, Yudai Nagano, Jin Zhang, Ryuichiro Hataya, Carmelo Militello, Andrea Tangherloni, Marco S. Nobile, Claudio Ferretti, Daniela Besozzi, Maria Carla Gilardi, Salvatore Vitabile, Giancarlo Mauri, Hideki Nakayama, Paolo Cazzaniga

The following mixed scheme is used for training/testing: (i) training on either each individual dataset or multiple prostate MRI datasets and (ii) testing on all three datasets with all possible training/testing combinations.

Combining Noise-to-Image and Image-to-Image GANs: Brain MR Image Augmentation for Tumor Detection

no code implementations31 May 2019 Changhee Han, Leonardo Rundo, Ryosuke Araki, Yudai Nagano, Yujiro Furukawa, Giancarlo Mauri, Hideki Nakayama, Hideaki Hayashi

In this context, Generative Adversarial Networks (GANs) can synthesize realistic/diverse additional training images to fill the data lack in the real image distribution; researchers have improved classification by augmenting data with noise-to-image (e. g., random noise samples to diverse pathological images) or image-to-image GANs (e. g., a benign image to a malignant one).

General Classification Image Augmentation +2

Generating Diverse Translations with Sentence Codes

no code implementations ACL 2019 Raphael Shu, Hideki Nakayama, Kyunghyun Cho

In this work, we attempt to obtain diverse translations by using sentence codes to condition the sentence generation.

Machine Translation Sentence +1

Empirical Study of Easy and Hard Examples in CNN Training

no code implementations ICLR 2019 Ikki Kishida, Hideki Nakayama

In this work, we study the similarities of easy and hard examples respectively for different Convolutional Neural Network (CNN) architectures, assessing how those examples contribute to generalization.

Incorporating Horizontal Connections in Convolution by Spatial Shuffling

no code implementations ICLR 2020 Ikki Kishida, Hideki Nakayama

In the view of the regular convolution's RF, the outputs of neurons in lower layers with smaller RF are bundled to create neurons in higher layers with larger RF.

Bridging the gap between AI and Healthcare sides: towards developing clinically relevant AI-powered diagnosis systems

no code implementations12 Jan 2020 Changhee Han, Leonardo Rundo, Kohei Murao, Takafumi Nemoto, Hideki Nakayama

Then, a questionnaire survey for physicians evaluates our pathology-aware Generative Adversarial Network (GAN)-based image augmentation projects in terms of Data Augmentation and physician training.

Generative Adversarial Network Image Augmentation +1

Single Model Ensemble using Pseudo-Tags and Distinct Vectors

no code implementations ACL 2020 Ryosuke Kuwabara, Jun Suzuki, Hideki Nakayama

Model ensemble techniques often increase task performance in neural networks; however, they require increased time, memory, and management effort.

Management text-classification +1

A Visually-Grounded Parallel Corpus with Phrase-to-Region Linking

no code implementations LREC 2020 Hideki Nakayama, Akihiro Tamura, Takashi Ninomiya

To verify our dataset, we performed phrase localization experiments in both languages and investigated the effectiveness of our Japanese annotations as well as multilingual learning realized by our dataset.

Image Captioning Multimodal Machine Translation +1

Meta Approach to Data Augmentation Optimization

no code implementations14 Jun 2020 Ryuichiro Hataya, Jan Zdenek, Kazuki Yoshizoe, Hideki Nakayama

Data augmentation policies drastically improve the performance of image recognition tasks, especially when the policies are optimized for the target data and tasks.

Classification Data Augmentation +2

MADGAN: unsupervised Medical Anomaly Detection GAN using multiple adjacent brain MRI slice reconstruction

no code implementations24 Jul 2020 Changhee Han, Leonardo Rundo, Kohei Murao, Tomoyuki Noguchi, Yuki Shimahara, Zoltan Adam Milacski, Saori Koshino, Evis Sala, Hideki Nakayama, Shinichi Satoh

Therefore, we propose unsupervised Medical Anomaly Detection Generative Adversarial Network (MADGAN), a novel two-step method using GAN-based multiple adjacent brain MRI slice reconstruction to detect brain anomalies at different stages on multi-sequence structural MRI: (Reconstruction) Wasserstein loss with Gradient Penalty + 100 L1 loss-trained on 3 healthy brain axial MRI slices to reconstruct the next 3 ones-reconstructs unseen healthy/abnormal scans; (Diagnosis) Average L2 loss per scan discriminates them, comparing the ground truth/reconstructed slices.

Generative Adversarial Network MRI Reconstruction +1

Graph-based Heuristic Search for Module Selection Procedure in Neural Module Network

no code implementations30 Sep 2020 Yuxuan Wu, Hideki Nakayama

To overcome this problem, existing work either included ground-truth program into training data or applied reinforcement learning to explore the program.

Question Answering reinforcement-learning +2

DJMix: Unsupervised Task-agnostic Augmentation for Improving Robustness

no code implementations1 Jan 2021 Ryuichiro Hataya, Hideki Nakayama

Convolutional Neural Networks (CNNs) are vulnerable to unseen noise on input images at the test time, and thus improving the robustness is crucial.

Data Augmentation Semantic Segmentation

A Visually-grounded First-person Dialogue Dataset with Verbal and Non-verbal Responses

no code implementations EMNLP 2020 Hisashi Kamezawa, Noriki Nishida, Nobuyuki Shimizu, Takashi Miyazaki, Hideki Nakayama

The results demonstrate that first-person vision helps neural network models correctly understand human intentions, and the production of non-verbal responses is a challenging task like that of verbal responses.

GraphPlan: Story Generation by Planning with Event Graph

no code implementations INLG (ACL) 2021 Hong Chen, Raphael Shu, Hiroya Takamura, Hideki Nakayama

In this paper, we focus on planning a sequence of events assisted by event graphs, and use the events to guide the generator.

Story Generation

Commonsense Knowledge Aware Concept Selection For Diverse and Informative Visual Storytelling

no code implementations5 Feb 2021 Hong Chen, Yifei HUANG, Hiroya Takamura, Hideki Nakayama

To enrich the candidate concepts, a commonsense knowledge graph is created for each image sequence from which the concept candidates are proposed.

Informativeness Visual Storytelling

Graph Energy-based Model for Substructure Preserving Molecular Design

no code implementations9 Feb 2021 Ryuichiro Hataya, Hideki Nakayama, Kazuki Yoshizoe

It is common practice for chemists to search chemical databases based on substructures of compounds for finding molecules with desired properties.

Supervised Visual Attention for Multimodal Neural Machine Translation

no code implementations COLING 2020 Tetsuro Nishihara, Akihiro Tamura, Takashi Ninomiya, Yutaro Omote, Hideki Nakayama

This paper proposed a supervised visual attention mechanism for multimodal neural machine translation (MNMT), trained with constraints based on manual alignments between words in a sentence and their corresponding regions of an image.

Machine Translation Sentence +1

Gradient-based Hyperparameter Optimization without Validation Data for Learning fom Limited Labels

no code implementations29 Sep 2021 Ryuichiro Hataya, Hideki Nakayama

Optimizing hyperparameters of machine learning algorithms especially for limited labeled data is important but difficult, because then obtaining enough validation data is practically impossible.

Hyperparameter Optimization Model Selection

SciXGen: A Scientific Paper Dataset for Context-Aware Text Generation

no code implementations Findings (EMNLP) 2021 Hong Chen, Hiroya Takamura, Hideki Nakayama

Generating texts in scientific papers requires not only capturing the content contained within the given input but also frequently acquiring the external information called \textit{context}.

Text Generation

Discrete Structural Planning for Generating Diverse Translations

no code implementations27 Sep 2018 Raphael Shu, Hideki Nakayama

Planning is important for humans when producing complex languages, which is a missing part in current language generation models.

Machine Translation Sentence +2

Unifying semi-supervised and robust learning by mixup

no code implementations ICLR Workshop LLD 2019 Ryuichiro Hataya, Hideki Nakayama

In this study, we consider learning from bi-quality data as a generalization of these studies, in which a small portion of data is cleanly labeled, and the rest is corrupt.

Learning with noisy labels

PPCD-GAN: Progressive Pruning and Class-Aware Distillation for Large-Scale Conditional GANs Compression

no code implementations16 Mar 2022 Duc Minh Vo, Akihiro Sugimoto, Hideki Nakayama

We push forward neural network compression research by exploiting a novel challenging task of large-scale conditional generative adversarial networks (GANs) compression.

Neural Network Compression

NOC-REK: Novel Object Captioning with Retrieved Vocabulary from External Knowledge

no code implementations CVPR 2022 Duc Minh Vo, Hong Chen, Akihiro Sugimoto, Hideki Nakayama

We propose an end-to-end Novel Object Captioning with Retrieved vocabulary from External Knowledge method (NOC-REK), which simultaneously learns vocabulary retrieval and caption generation, successfully describing novel objects outside of the training dataset.

Object object-detection +2

How do people talk about images? A study on open-domain conversations with images.

no code implementations NAACL (ACL) 2022 Yi-Pei Chen, Nobuyuki Shimizu, Takashi Miyazaki, Hideki Nakayama

This paper explores how humans conduct conversations with images by investigating an open-domain image conversation dataset, ImageChat.

Balancing Reconstruction and Editing Quality of GAN Inversion for Real Image Editing with StyleGAN Prior Latent Space

no code implementations31 May 2023 Kai Katsumata, Duc Minh Vo, Bei Liu, Hideki Nakayama

The exploration of the latent space in StyleGANs and GAN inversion exemplify impressive real-world image editing, yet the trade-off between reconstruction quality and editing quality remains an open problem.

Soft Curriculum for Learning Conditional GANs with Noisy-Labeled and Uncurated Unlabeled Data

no code implementations17 Jul 2023 Kai Katsumata, Duc Minh Vo, Tatsuya Harada, Hideki Nakayama

Label-noise or curated unlabeled data is used to compensate for the assumption of clean labeled data in training the conditional generative adversarial network; however, satisfying such an extended assumption is occasionally laborious or impractical.

Conditional Image Generation Generative Adversarial Network

Revisiting Latent Space of GAN Inversion for Real Image Editing

no code implementations18 Jul 2023 Kai Katsumata, Duc Minh Vo, Bei Liu, Hideki Nakayama

The exploration of the latent space in StyleGANs and GAN inversion exemplify impressive real-world image editing, yet the trade-off between reconstruction quality and editing quality remains an open problem.

EVCap: Retrieval-Augmented Image Captioning with External Visual-Name Memory for Open-World Comprehension

no code implementations27 Nov 2023 Jiaxuan Li, Duc Minh Vo, Akihiro Sugimoto, Hideki Nakayama

Large language models (LLMs)-based image captioning has the capability of describing objects not explicitly observed in training data; yet novel objects occur frequently, necessitating the requirement of sustaining up-to-date object knowledge for open-world comprehension.

Image Captioning Object +1

LayoutFlow: Flow Matching for Layout Generation

no code implementations27 Mar 2024 Julian Jorge Andrade Guerreiro, Naoto Inoue, Kento Masui, Mayu Otani, Hideki Nakayama

Finding a suitable layout represents a crucial task for diverse applications in graphic design.

Cannot find the paper you are looking for? You can Submit a new open access paper.