Search

Contextualized word embeddings can lead to state-of-the-art performances in natural language understanding. Recently, a pre-trained deep contextualized text encoder such as BERT has shown its potential in improving natural language tasks including abstractive summarization. Existing approaches in dialogue summarization focus on incorporating a large language model into summarization task trained on large-scale corpora consisting of news articles rather than dialogues of multiple speakers. In this paper, we introduce self-supervised methods to compensate shortcomings to train a dialogue summarization model. Our principle is to detect incoherent information flows using pretext dialogue text to enhance BERT's ability to contextualize the dialogue text representations. We build and fine-tune an abstractive dialogue summarization model on a shared encoder-decoder architecture using the enhanced BERT. We empirically evaluate our abstractive dialogue summarizer with the SAMSum corpus, a recently introduced dataset with abstractive dialogue summaries. All of our methods have contributed improvements to abstractive summary measured in ROUGE scores. Through an extensive ablation study, we also present a sensitivity analysis to critical model hyperparameters, probabilities of switching utterances and masking interlocutors., 5 pages, 3 figures, INTERSPEECH 2021

The use of electrocardiogram (ECG) data for personal identification in Industrial Internet of Things can achieve near-perfect accuracy in an ideal condition. However, real-life ECG data are often exposed to various types of noises and interferences. A reliable and enhanced identification method could be achieved by employing additional features from other biometric sources. This work, thus, proposes a novel robust and reliable identification technique grounded on multimodal biometrics, which utilizes deep learning to combine fingerprint, ECG and facial image data, particularly useful for identification and gender classification purposes. The multimodal approach allows the model to deal with a range of input domains removing the requirement of independent training on each modality, and inter-domain correlation can improve the model generalization capability on these tasks. In multitask learning, losses from one task help to regularize others, thus, leading to better overall performances. The proposed approach merges the embedding of multimodality by using feature-level and score level fusions. To the best of our understanding, the key concepts presented herein is a pioneering work combining multimodality, multitasking and different fusion methods. The proposed model achieves a better generalization on the benchmark dataset used while the feature-level fusion outperforms other fusion methods. The proposed model is validated on noisy and incomplete data with missing modalities and the analyses on the experimental results are provided.

Background Recognition is an essential function of human beings. Humans easily recognize a person using various inputs such as voice, face, or gesture. In this study, we mainly focus on DL model with multi-modality which has many benefits including noise reduction. We used ResNet-50 for extracting features from dataset with 2D data. Results This study proposes a novel multimodal and multitask model, which can both identify human ID and classify the gender in single step. At the feature level, the extracted features are concatenated as the input for the identification module. Additionally, in our model design, we can change the number of modalities used in a single model. To demonstrate our model, we generate 58 virtual subjects with public ECG, face and fingerprint dataset. Through the test with noisy input, using multimodal is more robust and better than using single modality. Conclusions This paper presents an end-to-end approach for multimodal and multitask learning. The proposed model shows robustness on the spoof attack, which can be significant for bio-authentication device. Through results in this study, we suggest a new perspective for human identification task, which performs better than in previous approaches.

Autonomous driving technologies (ADTs) are transformative because they are expected to assume the task of driving. Since travel distance is closely related to the burden of driving, it is expected to affect the ADT preference; however, this effect has not been investigated in the transportation literature. Therefore, this study is designed to investigate the ADT preference and heterogeneity with respect to different travel distances. Hypothetical choice situations were designed to investigate people's ADT preferences over human driving for varying travel distances. Using the stated preference data collected via the experiments, four latent class models were estimated for heterogeneity across classes and travel distances. The estimation results show that three classes, we labeled as competitive, autonomous vehicle oriented, human-driven vehicle oriented, are revealed by the ADT preference, regardless of the travel distance. In addition, in-vehicle activities in hypothetical autonomous driving situations were observed across all the classes and travel distances.

An amendment to this paper has been published and can be accessed via the original article.

The demands on higher bandwidth with reduced power consumption in mobile market are driving mobile DRAM with advanced design techniques. Proposed LPDDR4 in this paper achieves over 39% improvement in power efficiency and over 4.3 Gbps data rate with 1.1 V supply voltage. These are challenging targets compared with those of LPDDR3. This work describes design schemes employed in LPDDR4 to satisfy these requirements, such as multi-channel-per-die architecture, multiple training modes, low-swing interface, DQS and clock frequency dividing, and internal reference for data and command-address signals. This chip was fabricated in a 3-metal 2y-nm DRAM CMOS process.