Search

In this paper, we propose a deep learning based system for the task of deepfake audio detection. In particular, the draw input audio is first transformed into various spectrograms using three transformation methods of Short-time Fourier Transform (STFT), Constant-Q Transform (CQT), Wavelet Transform (WT) combined with different auditory-based filters of Mel, Gammatone, linear filters (LF), and discrete cosine transform (DCT). Given the spectrograms, we evaluate a wide range of classification models based on three deep learning approaches. The first approach is to train directly the spectrograms using our proposed baseline models of CNN-based model (CNN-baseline), RNN-based model (RNN-baseline), C-RNN model (C-RNN baseline). Meanwhile, the second approach is transfer learning from computer vision models such as ResNet-18, MobileNet-V3, EfficientNet-B0, DenseNet-121, SuffleNet-V2, Swint, Convnext-Tiny, GoogLeNet, MNASsnet, RegNet. In the third approach, we leverage the state-of-the-art audio pre-trained models of Whisper, Seamless, Speechbrain, and Pyannote to extract audio embeddings from the input spectrograms. Then, the audio embeddings are explored by a Multilayer perceptron (MLP) model to detect the fake or real audio samples. Finally, high-performance deep learning models from these approaches are fused to achieve the best performance. We evaluated our proposed models on ASVspoof 2019 benchmark dataset. Our best ensemble model achieved an Equal Error Rate (EER) of 0.03, which is highly competitive to top-performing systems in the ASVspoofing 2019 challenge. Experimental results also highlight the potential of selective spectrograms and deep learning approaches to enhance the task of audio deepfake detection.

Evaluating text summarization has been a challenging task in natural language processing (NLP). Automatic metrics which heavily rely on reference summaries are not suitable in many situations, while human evaluation is time-consuming and labor-intensive. To bridge this gap, this paper proposes a novel method based on large language models (LLMs) for evaluating text summarization. We also conducts a comparative study on eight automatic metrics, human evaluation, and our proposed LLM-based method. Seven different types of state-of-the-art (SOTA) summarization models were evaluated. We perform extensive experiments and analysis on datasets with patent documents. Our results show that LLMs evaluation aligns closely with human evaluation, while widely-used automatic metrics such as ROUGE-2, BERTScore, and SummaC do not and also lack consistency. Based on the empirical comparison, we propose a LLM-powered framework for automatically evaluating and improving text summarization, which is beneficial and could attract wide attention among the community., Comment: The paper is under review at Empirical Methods in Natural Language Processing (EMNLP) 2024. It has 15 pages and 4 figures

Understanding chemical modifications on proteins opens up further possibilities for research on rare diseases. This work proposes visualization approaches using two-dimensional (2D) and three-dimensional (3D) visual representations to analyze and gain insights into protein modifications. In this work, we present the application of Modie Viewer as an attempt to address the Bio+MedVis Challenge at IEEE VIS 2022., Comment: 5 pages, 5 figures, Bio+MedVis Challenge @ IEEE VIS 2022

Whether it is in the form of transcribed conversations, blog posts, or tweets, qualitative data provides a reader with rich insight into both the overarching trends as well as the diversity of human ideas expressed through text. Handling and analyzing large amounts of qualitative data, however, is difficult, often requiring multiple time-intensive perusals in order to identify patterns. This difficulty is multiplied with each additional question or time point present in a data set. A primary challenge then is creating visualizations that support the interpretation of qualitative data by making it easier to identify and explore trends of interest. By combining the affordances of both text and visualizations, WordStream has previously enabled ease of information retrieval and processing of time-series text data, but the data-wrangling necessary to produce a WordStream remains a significant barrier for non-technical users. In response, this paper presents WordStream Maker: an end-to-end platform with a pipeline that utilizes natural language processing (NLP) to help non-technical users process raw text data and generate a customizable visualization without programming practice. Lessons learned from integrating NLP into visualization and scaling to large data sets are discussed, along with use cases to demonstrate the usefulness of the platform., Comment: 4 pages, 4 figures. In NLVIZ: Exploring Research Opportunities for Natural Language, Text, and Data Visualization, IEEE VIS 2022

Image augmentation techniques have been widely investigated to improve the performance of deep learning (DL) algorithms on mammography classification tasks. Recent methods have proved the efficiency of image augmentation on data deficiency or data imbalance issues. In this paper, we propose a novel transparency strategy to boost the Breast Imaging Reporting and Data System (BI-RADS) scores of mammogram classifiers. The proposed approach utilizes the Region of Interest (ROI) information to generate more high-risk training examples for breast cancer (BI-RADS 3, 4, 5) from original images. Our extensive experiments on three different datasets show that the proposed approach significantly improves the mammogram classification performance and surpasses a state-of-the-art data augmentation technique called CutMix. This study also highlights that our transparency method is more effective than other augmentation strategies for BI-RADS classification and can be widely applied to other computer vision tasks., Comment: Accepted for presentation at the 22nd IEEE Statistical Signal Processing (SSP) workshop

Advanced deep learning (DL) algorithms may predict the patient's risk of developing breast cancer based on the Breast Imaging Reporting and Data System (BI-RADS) and density standards. Recent studies have suggested that the combination of multi-view analysis improved the overall breast exam classification. In this paper, we propose a novel multi-view DL approach for BI-RADS and density assessment of mammograms. The proposed approach first deploys deep convolutional networks for feature extraction on each view separately. The extracted features are then stacked and fed into a Light Gradient Boosting Machine (LightGBM) classifier to predict BI-RADS and density scores. We conduct extensive experiments on both the internal mammography dataset and the public dataset Digital Database for Screening Mammography (DDSM). The experimental results demonstrate that the proposed approach outperforms the single-view classification approach on two benchmark datasets by huge F1-score margins (+5% on the internal dataset and +10% on the DDSM dataset). These results highlight the vital role of combining multi-view information to improve the performance of breast cancer risk prediction., Comment: This paper has been accepted by the 44th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (2022 IEEE EMBC)

This paper presents VisMCA, an interactive visual analytics system that supports deepening understanding in ML results, augmenting users' capabilities in correcting misclassification, and providing an analysis of underlying patterns, in response to the VAST Challenge 2020 Mini-Challenge 2. VisMCA facilitates tracking provenance and provides a comprehensive view of object detection results, easing re-labeling, and producing reliable, corrected data for future training. Our solution implements multiple analytical views on visual analysis to offer a deep insight for underlying pattern discovery.

This paper reports on the ReINTEL Shared Task for Responsible Information Identification on social network sites, which is hosted at the seventh annual workshop on Vietnamese Language and Speech Processing (VLSP 2020). Given a piece of news with respective textual, visual content and metadata, participants are required to classify whether the news is `reliable' or `unreliable'. In order to generate a fair benchmark, we introduce a novel human-annotated dataset of over 10,000 news collected from a social network in Vietnam. All models will be evaluated in terms of AUC-ROC score, a typical evaluation metric for classification. The competition was run on the Codalab platform. Within two months, the challenge has attracted over 60 participants and recorded nearly 1,000 submission entries.

Having a good Human-Computer Interaction (HCI) design is challenging. Previous works have contributed significantly to fostering HCI, including design principle with report study from the instructor view. The questions of how and to what extent students perceive the design principles are still left open. To answer this question, this paper conducts a study of HCI adoption in the classroom. The studio-based learning method was adapted to teach 83 graduate and undergraduate students in 16 weeks long with four activities. A standalone presentation tool for instant online peer feedback during the presentation session was developed to help students justify and critique other's work. Our tool provides a sandbox, which supports multiple application types, including Web-applications, Object Detection, Web-based Virtual Reality (VR), and Augmented Reality (AR). After presenting one assignment and two projects, our results showed that students acquired a better understanding of the Golden Rules principle over time, which was demonstrated by the development of visual interface design. The Wordcloud reveals the primary focus was on the user interface and shed some light on students' interest in user experience. The inter-rater score indicates the agreement among students that they have the same level of understanding of the principles. The results show a high level of guideline compliance with HCI principles, in which we witnessed variations in visual cognitive styles. Regardless of diversity in visual preference, the students presented high consistency and a similar perspective on adopting HCI design principles. The results also elicited suggestions into the development of the HCI curriculum in the future., Comment: 12 pages, 4 figures, 15th International Symposium on Visual Computing 2020

Citation function and citation sentiment are two essential aspects of citation content analysis (CCA), which are useful for influence analysis, the recommendation of scientific publications. However, existing studies are mostly traditional machine learning methods, although deep learning techniques have also been explored, the improvement of the performance seems not significant due to insufficient training data, which brings difficulties to applications. In this paper, we propose to fine-tune pre-trained contextual embeddings ULMFiT, BERT, and XLNet for the task. Experiments on three public datasets show that our strategy outperforms all the baselines in terms of the F1 score. For citation function identification, the XLNet model achieves 87.2%, 86.90%, and 81.6% on DFKI, UMICH, and TKDE2019 datasets respectively, while it achieves 91.72% and 91.56% on DFKI and UMICH in term of citation sentiment identification. Our method can be used to enhance the influence analysis of scholars and scholarly publications., Comment: 1 figure and three tables

Entrepreneurs play crucial roles in global sustainable development, but limited financial resources constrain their performance and survival rate. Entrepreneurial finance discipline is, therefore, born to explore the connection between finance and entrepreneurship. Despite the global presence of entrepreneurship, the literature of entrepreneurial finance is suspected to be Western ideologically homogenous. Thus, the objective of this study is to examine the existence of Western ideological homogeneity in entrepreneurial finance literature. Employing the mindsponge mechanism and bibliometric analyses (Y-index and social structure), we analyze 412 highly cited publications extracted from Web of Science database and find Western ideological dominance as well as weak tolerance towards heterogeneity in the set of core ideologies of entrepreneurial finance. These results are consistent across author-, institution-, and country-levels, which reveals strong evidence for the existence of Western ideological homogeneity in the field. We recommend editors, reviewers, and authors to have proactive actions to diversify research topics and enhancing knowledge exchange to avoid the shortfalls of ideological homogeneity. Moreover, the synthesis of mindsponge mechanism and bibliometric analyses are suggested as a possible way to evaluate the state of ideological diversity in other scientific disciplines., Comment: 27 pages, 8 figures

The paper describes the organisation of the "HateSpeech Detection" (HSD) task at the VLSP workshop 2019 on detecting the fine-grained presence of hate speech in Vietnamese textual items (i.e., messages) extracted from Facebook, which is the most popular social network site (SNS) in Vietnam. The task is organised as a multi-class classification task and based on a large-scale dataset containing 25,431 Vietnamese textual items from Facebook. The task participants were challenged to build a classification model that is capable of classifying an item to one of 3 classes, i.e., "HATE", "OFFENSIVE" and "CLEAN". HSD attracted a large number of participants and was a popular task at VLSP 2019. In particular, there were 71 teams signed up for the task, 14 of them submitted results with 380 valid submissions from 20th September 2019 to 4th October 2019.

With language modeling becoming the popular base task for unsupervised representation learning in Natural Language Processing, it is important to come up with new architectures and techniques for faster and better training of language models. However, due to a peculiarity of languages -- the larger the dataset, the higher the average number of times a word appears in that dataset -- datasets of different sizes have very different properties. Architectures performing well on small datasets might not perform well on larger ones. For example, LSTM models perform well on WikiText-2 but poorly on WikiText-103, while Transformer models perform well on WikiText-103 but not on WikiText-2. For setups like architectural search, this is a challenge since it is prohibitively costly to run a search on the full dataset but it is not indicative to experiment on smaller ones. In this paper, we introduce SimpleBooks, a small dataset with the average word frequency as high as that of much larger ones. Created from 1,573 Gutenberg books with the highest ratio of word-level book length to vocabulary size, SimpleBooks contains 92M word-level tokens, on par with WikiText-103 (103M tokens), but has the vocabulary of 98K, a third of WikiText-103's. SimpleBooks can be downloaded from https://dldata-public.s3.us-east-2.amazonaws.com/simplebooks.zip., Comment: 5 pages

This paper introduces EQSA, an interactive exploratory tool for earthquake situational analytics using social media. EQSA is designed to support users to characterize the condition across the area around the earthquake zone, regarding related events, resources to be allocated, and responses from the community. On the general level, changes in the volume of messages from chosen categories are presented, assisting users in conveying a general idea of the condition. More in-depth analysis is provided with topic evolution, community visualization, and location representation. EQSA is developed with intuitive, interactive features and multiple linked views, visualizing social media data, and supporting users to gain a comprehensive insight into the situation. In this paper, we present the application of EQSA with the VAST Challenge 2019: Mini-Challenge 3 (MC3) dataset., Comment: 2 pages, 2 figures

NeMo (Neural Modules) is a Python framework-agnostic toolkit for creating AI applications through re-usability, abstraction, and composition. NeMo is built around neural modules, conceptual blocks of neural networks that take typed inputs and produce typed outputs. Such modules typically represent data layers, encoders, decoders, language models, loss functions, or methods of combining activations. NeMo makes it easy to combine and re-use these building blocks while providing a level of semantic correctness checking via its neural type system. The toolkit comes with extendable collections of pre-built modules for automatic speech recognition and natural language processing. Furthermore, NeMo provides built-in support for distributed training and mixed precision on latest NVIDIA GPUs. NeMo is open-source https://github.com/NVIDIA/NeMo, Comment: 6 pages plus references

We propose NovoGrad, an adaptive stochastic gradient descent method with layer-wise gradient normalization and decoupled weight decay. In our experiments on neural networks for image classification, speech recognition, machine translation, and language modeling, it performs on par or better than well tuned SGD with momentum and Adam or AdamW. Additionally, NovoGrad (1) is robust to the choice of learning rate and weight initialization, (2) works well in a large batch setting, and (3) has two times smaller memory footprint than Adam., Comment: Preprint, under review

In this paper, we report state-of-the-art results on LibriSpeech among end-to-end speech recognition models without any external training data. Our model, Jasper, uses only 1D convolutions, batch normalization, ReLU, dropout, and residual connections. To improve training, we further introduce a new layer-wise optimizer called NovoGrad. Through experiments, we demonstrate that the proposed deep architecture performs as well or better than more complex choices. Our deepest Jasper variant uses 54 convolutional layers. With this architecture, we achieve 2.95% WER using a beam-search decoder with an external neural language model and 3.86% WER with a greedy decoder on LibriSpeech test-clean. We also report competitive results on the Wall Street Journal and the Hub5'00 conversational evaluation datasets., Comment: Accepted to INTERSPEECH 2019

We present OpenSeq2Seq - a TensorFlow-based toolkit for training sequence-to-sequence models that features distributed and mixed-precision training. Benchmarks on machine translation and speech recognition tasks show that models built using OpenSeq2Seq give state-of-the-art performance at 1.5-3x less training time. OpenSeq2Seq currently provides building blocks for models that solve a wide range of tasks including neural machine translation, automatic speech recognition, and speech synthesis., Comment: Presented at Workshop for Natural Language Processing Open Source Software (NLP-OSS), co-located with ACL2018

A dynamic partial order reduction (DPOR) algorithm is optimal when it always explores at most one representative per Mazurkiewicz trace. Existing literature suggests that the reduction obtained by the non-optimal, state-of-the-art Source-DPOR (SDPOR) algorithm is comparable to optimal DPOR. We show the first program with $\mathop{\mathcal{O}}(n)$ Mazurkiewicz traces where SDPOR explores $\mathop{\mathcal{O}}(2^n)$ redundant schedules and identify the cause of the blow-up as an NP-hard problem. Our main contribution is a new approach, called Quasi-Optimal POR, that can arbitrarily approximate an optimal exploration using a provided constant k. We present an implementation of our method in a new tool called Dpu using specialised data structures. Experiments with Dpu, including Debian packages, show that optimality is achieved with low values of k, outperforming state-of-the-art tools., Comment: Minor corrections after review for publication