Wenxuan Wang, Wenxiang Jiao, Yongchang Hao, Xing Wang, Shuming Shi, Zhaopeng Tu, Michael Lyu
"In this paper, we present a substantial step in better understanding the SOTA sequence-to-sequence (Seq2Seq) pretraining for neural machine translation~(NMT). We focus on studying the impact of the jointly pretrained decoder, which is the main difference between Seq2Seq pretraining and previous encoder-based pretraining approaches for NMT. By carefully designing experiments on three language pairs, we find that Seq2Seq pretraining is a double-edged sword: On one hand, it helps NMT models to produce more diverse translations and reduce adequacy-related translation errors. On the other hand, the discrepancies between Seq2Seq pretraining and NMT finetuning limit the translation quality (i.e., domain discrepancy) and induce the over-estimation issue (i.e., objective discrepancy). Based on these observations, we further propose simple and effective strategies, named in-domain pretraining and input adaptation to remedy the domain and objective discrepancies, respectively. Experimental results on several language pairs show that our approach can consistently improve both translation performance and model robustness upon Seq2Seq pretraining."
https://arxiv.org/abs/2203.08442
Jianfeng Wang, Zhengyuan Yang, Xiaowei Hu, Linjie Li, Kevin Lin, Zhe Gan, Zicheng Liu, Ce Liu, Lijuan Wang
"In this paper, we design and train a Generative Image-to-text Transformer, GIT, to unify vision-language tasks such as image/video captioning and question answering. While generative models provide a consistent network architecture between pre-training and fine-tuning, existing work typically contains complex structures (uni/multi-modal encoder/decoder) and depends on external modules such as object detectors/taggers and optical character recognition (OCR). In GIT, we simplify the architecture as one image encoder and one text decoder under a single language modeling task. We also scale up the pre-training data and the model size to boost the model performance. Without bells and whistles, our GIT establishes new state of the arts on 12 challenging benchmarks with a large margin. For instance, our model surpasses the human performance for the first time on TextCaps (138.2 vs. 125.5 in CIDEr). Furthermore, we present a new scheme of generation-based image classification and scene text recognition, achieving decent performance on standard benchmarks."
https://arxiv.org/abs/2205.14100
Sascha Rothe, Shashi Narayan, Aliaksei Severyn
"Unsupervised pre-training of large neural models has recently revolutionized Natural Language Processing. By warm-starting from the publicly released checkpoints, NLP practitioners have pushed the state-of-the-art on multiple benchmarks while saving significant amounts of compute time. So far the focus has been mainly on the Natural Language Understanding tasks. In this paper, we demonstrate the efficacy of pre-trained checkpoints for Sequence Generation. We developed a Transformer-based sequence-to-sequence model that is compatible with publicly available pre-trained BERT, GPT-2 and RoBERTa checkpoints and conducted an extensive empirical study on the utility of initializing our model, both encoder and decoder, with these checkpoints. Our models result in new state-of-the-art results on Machine Translation, Text Summarization, Sentence Splitting, and Sentence Fusion."
https://arxiv.org/abs/1907.12461
Felix Wu, Kwangyoun Kim, Shinji Watanabe, Kyu Han, Ryan McDonald, Kilian Q. Weinberger, Yoav Artzi
"We introduce Wav2Seq, the first self-supervised approach to pre-train both parts of encoder-decoder models for speech data. We induce a pseudo language as a compact discrete representation, and formulate a self-supervised pseudo speech recognition task -- transcribing audio inputs into pseudo subword sequences. This process stands on its own, or can be applied as low-cost second-stage pre-training. We experiment with automatic speech recognition (ASR), spoken named entity recognition, and speech-to-text translation. We set new state-of-the-art results for end-to-end spoken named entity recognition, and show consistent improvements on 20 language pairs for speech-to-text translation, even when competing methods use additional text data for training. Finally, on ASR, our approach enables encoder-decoder methods to benefit from pre-training for all parts of the network, and shows comparable performance to highly optimized recent methods."
https://arxiv.org/abs/2205.01086
William Chan, Navdeep Jaitly, Quoc V. Le, Oriol Vinyals
"We present Listen, Attend and Spell (LAS), a neural network that learns to transcribe speech utterances to characters. Unlike traditional DNN-HMM models, this model learns all the components of a speech recognizer jointly. Our system has two components: a listener and a speller. The listener is a pyramidal recurrent network encoder that accepts filter bank spectra as inputs. The speller is an attention-based recurrent network decoder that emits characters as outputs. The network produces character sequences without making any independence assumptions between the characters. This is the key improvement of LAS over previous end-to-end CTC models. On a subset of the Google voice search task, LAS achieves a word error rate (WER) of 14.1% without a dictionary or a language model, and 10.3% with language model rescoring over the top 32 beams. By comparison, the state-of-the-art CLDNN-HMM model achieves a WER of 8.0%."
https://arxiv.org/abs/1508.01211
Changhan Wang, Anne Wu, Juan Pino, Alexei Baevski, Michael Auli, Alexis Conneau
"In this paper, we improve speech translation (ST) through effectively leveraging large quantities of unlabeled speech and text data in different and complementary ways. We explore both pretraining and self-training by using the large Libri-Light speech audio corpus and language modeling with CommonCrawl. Our experiments improve over the previous state of the art by 2.6 BLEU on average on all four considered CoVoST 2 language pairs via a simple recipe of combining wav2vec 2.0 pretraining, a single iteration of self-training and decoding with a language model. Different to existing work, our approach does not leverage any other supervision than ST data. Code and models will be publicly released."
https://arxiv.org/abs/2104.06678
Junyi Ao, Ziqiang Zhang, Long Zhou, Shujie Liu, Haizhou Li, Tom Ko, Lirong Dai, Jinyu Li, Yao Qian, Furu Wei
"This paper studies a novel pre-training technique with unpaired speech data, Speech2C, for encoder-decoder based automatic speech recognition (ASR). Within a multi-task learning framework, we introduce two pre-training tasks for the encoder-decoder network using acoustic units, i.e., pseudo codes, derived from an offline clustering model. One is to predict the pseudo codes via masked language modeling in encoder output, like HuBERT model, while the other lets the decoder learn to reconstruct pseudo codes autoregressively instead of generating textual scripts. In this way, the decoder learns to reconstruct original speech information with codes before learning to generate correct text. Comprehensive experiments on the LibriSpeech corpus show that the proposed Speech2C can relatively reduce the word error rate (WER) by 19.2% over the method without decoder pre-training, and also outperforms significantly the state-of-the-art wav2vec 2.0 and HuBERT on fine-tuning subsets of 10h and 100h."
https://arxiv.org/abs/2203.17113
Chitwan Saharia, Jonathan Ho, William Chan, Tim Salimans, David J. Fleet, Mohammad Norouzi
"We present SR3, an approach to image Super-Resolution via Repeated Refinement. SR3 adapts denoising diffusion probabilistic models to conditional image generation and performs super-resolution through a stochastic denoising process. Inference starts with pure Gaussian noise and iteratively refines the noisy output using a U-Net model trained on denoising at various noise levels. SR3 exhibits strong performance on super-resolution tasks at different magnification factors, on faces and natural images. We conduct human evaluation on a standard 8X face super-resolution task on CelebA-HQ, comparing with SOTA GAN methods. SR3 achieves a fool rate close to 50%, suggesting photo-realistic outputs, while GANs do not exceed a fool rate of 34%. We further show the effectiveness of SR3 in cascaded image generation, where generative models are chained with super-resolution models, yielding a competitive FID score of 11.3 on ImageNet."
https://arxiv.org/abs/2104.07636
WEIFENG GE, BINGCHEN GONG, and YIZHOU YU, The University of Hong Kong
"Single image superresolution has been a popular research topic in the last two decades and has recently received a new wave of interest due to deep neural networks. In this paper, we approach this problem from a different perspective. With respect to a downsampled low resolution image, we model a high resolution image as a combination of two components, a deterministic component and a stochastic component. The deterministic component can be recovered from the low-frequency signals in the downsampled image. The stochastic component, on the other hand, contains the signals that have little correlation with the low resolution image. We adopt two complementary methods for generating these two components. While generative adversarial networks are used for the stochastic component, deterministic component reconstruction is formulated as a regression problem solved using deep neural networks. Since the deterministic component exhibits clearer local orientations, we design novel loss functions tailored for such properties for training the deep regression network. These two methods are first applied to the entire input image to produce two distinct high-resolution images. Afterwards, these two images are fused together using another deep neural network that also performs local statistical rectification, which tries to make the local statistics of the fused image match the same local statistics of the groundtruth image. Quantitative results and a user study indicate that the proposed method outperforms existing state-of-the-art algorithms with a clear margin. "
https://arxiv.org/pdf/1809.06557.pdf
Curriculum Learning: A Regularization Method for Efficient and Stable Billion-Scale GPT Model Pre-Training
Conglong Li, Minjia Zhang, Yuxiong He
"Recent works have demonstrated great success in training high-capacity autoregressive language models (GPT, GPT-2, GPT-3) on a huge amount of unlabeled text corpus for text generation. Despite showing great results, this generates two training efficiency challenges. First, training large corpora can be extremely timing consuming, and how to present training samples to the model to improve the token-wise convergence speed remains a challenging and open question. Second, many of these large models have to be trained with hundreds or even thousands of processors using data-parallelism with a very large batch size. Despite of its better compute efficiency, it has been observed that large-batch training often runs into training instability issue or converges to solutions with bad generalization performance. To overcome these two challenges, we present a study of a curriculum learning based approach, which helps improves the pre-training convergence speed of autoregressive models. More importantly, we find that curriculum learning, as a regularization method, exerts a gradient variance reduction effect and enables to train autoregressive models with much larger batch sizes and learning rates without training instability, further improving the training speed. Our evaluations demonstrate that curriculum learning enables training GPT-2 models (with up to 1.5B parameters) with 8x larger batch size and 4x larger learning rate, whereas the baseline approach struggles with training divergence. To achieve the same validation perplexity targets during pre-training, curriculum learning reduces the required number of tokens and wall clock time by up to 59% and 54%, respectively. To achieve the same or better zero-shot WikiText-103/LAMBADA evaluation results at the end of pre-training, curriculum learning reduces the required number of tokens and wall clock time by up to 13% and 61%, respectively. "
https://arxiv.org/abs/2108.06084
"Artificial Intelligence is developing fast. It will change our lives by improving healthcare (e.g. making diagnosis more precise, enabling better prevention of diseases), increasing the efficiency of farming, contributing to climate change mitigation and adaptation, improving the efficiency of production systems through predictive maintenance, increasing the security of Europeans, and in many other ways that we can only begin to imagine. At the same time, Artificial Intelligence (AI) entails a number of potential risks, such as opaque decision-making, gender-based or other kinds of discrimination, intrusion in our private lives or being used for criminal purposes."
Yeonsoo Lim, Deokjin Seo, and Yuchul Jung
"Despite extensive research, performance enhancement of keyphrase (KP) extractionremains a challenging problem in modern informatics. Recently, deep learning-based supervised approaches have exhibited state-of-the-art accuracies with respect to this problem, and several of the previously proposed methods utilize Bidirectional Encoder Representations from Transformers (BERT)-based language models. However, few studies have investigated the effective application of BERT based fine-tuning techniques to the problem of KP extraction. In this paper, we consider the aforementioned problem in the context of scientific articles by investigating the fine-tuning characteristics of two distinct BERT models — BERT (i.e., base BERT model by Google) and SciBERT (i.e., a BERT model trained on scientific text). Three different datasets (WWW, KDD, and Inspec) comprising data obtained from the computer science domain are used to compare the results obtained by fine-tuning BERT and SciBERT in terms of KP extraction"
http://jaitc.ki-it.or.kr/xml/24833/24833.pdf
Zejiang Shen, Ruochen Zhang, Melissa Dell, Benjamin Charles Germain Lee, Jacob Carlson, Weining Li
"Recent advances in document image analysis (DIA) have been primarily driven by the application of neural networks. Ideally, research outcomes could be easily deployed in production and extended for further investigation. However, various factors like loosely organized codebases and sophisticated model configurations complicate the easy reuse of important innovations by a wide audience. Though there have been on-going efforts to improve reusability and simplify deep learning (DL) model development in disciplines like natural language processing and computer vision, none of them are optimized for challenges in the domain of DIA. This represents a major gap in the existing toolkit, as DIA is central to academic research across a wide range of disciplines in the social sciences and humanities. This paper introduces layoutparser, an open-source library for streamlining the usage of DL in DIA research and applications. The core layoutparser library comes with a set of simple and intuitive interfaces for applying and customizing DL models for layout detection, character recognition, and many other document processing tasks. To promote extensibility, layoutparser also incorporates a community platform for sharing both pre-trained models and full document digitization pipelines. We demonstrate that layoutparser is helpful for both lightweight and large-scale digitization pipelines in real-word use cases. The library is publicly available at this https URL."
https://arxiv.org/abs/2103.15348
Oleksii Hrinchuk, Mariya Popova, Boris Ginsburg
"In this work, we introduce a simple yet efficient post-processing model for automatic speech recognition (ASR). Our model has Transformer-based encoder-decoder architecture which "translates" ASR model output into grammatically and semantically correct text. We investigate different strategies for regularizing and optimizing the model and show that extensive data augmentation and the initialization with pre-trained weights are required to achieve good performance. On the LibriSpeech benchmark, our method demonstrates significant improvement in word error rate over the baseline acoustic model with greedy decoding, especially on much noisier dev-other and test-other portions of the evaluation dataset. Our model also outperforms baseline with 6-gram language model re-scoring and approaches the performance of re-scoring with Transformer-XL neural language model."
https://arxiv.org/abs/1910.10697
Kazuki Irie, Rohit Prabhavalkar, Anjuli Kannan, Antoine Bruguier, David Rybach, Patrick Nguyen
"In conventional speech recognition, phoneme-based models outperform grapheme-based models for non-phonetic languages such as English. The performance gap between the two typically reduces as the amount of training data is increased. In this work, we examine the impact of the choice of modeling unit for attention-based encoder-decoder models. We conduct experiments on the LibriSpeech 100hr, 460hr, and 960hr tasks, using various target units (phoneme, grapheme, and word-piece); across all tasks, we find that grapheme or word-piece models consistently outperform phoneme-based models, even though they are evaluated without a lexicon or an external language model. We also investigate model complementarity: we find that we can improve WERs by up to 9% relative by rescoring N-best lists generated from a strong word-piece based baseline with either the phoneme or the grapheme model. Rescoring an N-best list generated by the phonemic system, however, provides limited improvements. Further analysis shows that the word-piece-based models produce more diverse N-best hypotheses, and thus lower oracle WERs, than phonemic models."
https://arxiv.org/abs/1902.01955
Junwei Liao, Sefik Emre Eskimez, Liyang Lu, Yu Shi, Ming Gong, Linjun Shou, Hong Qu, Michael Zeng,
"Modern Automatic Speech Recognition (ASR) systems can achieve high performance in terms of recognition accuracy. However, a perfectly accurate transcript still can be challenging to read due to grammatical errors, disfluency, and other errata common in spoken communication. Many downstream tasks and human readers rely on the output of the ASR system; therefore, errors introduced by the speaker and ASR system alike will be propagated to the next task in the pipeline. In this work, we propose a novel NLP task called ASR post-processing for readability (APR) that aims to transform the noisy ASR output into a readable text for humans and downstream tasks while maintaining the semantic meaning of the speaker. In addition, we describe a method to address the lack of task-specific data by synthesizing examples for the APR task using the datasets collected for Grammatical Error Correction (GEC) followed by text-to-speech (TTS) and ASR. Furthermore, we propose metrics borrowed from similar tasks to evaluate performance on the APR task. We compare fine-tuned models based on several open-sourced and adapted pre-trained models with the traditional pipeline method. Our results suggest that finetuned models improve the performance on the APR task significantly, hinting at the potential benefits of using APR systems. We hope that the read, understand, and rewrite approach of our work can serve as a basis that many NLP tasks and human readers can benefit from."
https://arxiv.org/pdf/2004.04438.pdf
Markus Näther
"Determining and correcting spelling and grammar errors in text is an important but surprisingly difficult task. There are several reasons why this remains challenging. Errors may consist of simple typing errors like deleted, substituted, or wrongly inserted letters, but may also consist of word confusions where a word was replaced by another one. In addition, words may be erroneously split into two parts or get concatenated. Some words can contain hyphens, because they were split at the end of a line or are compound words with a mandatory hyphen. In this paper, we provide an extensive evaluation of 14 spelling correction tools on a common benchmark. In particular, the evaluation provides a detailed comparison with respect to 12 error categories. The benchmark consists of sentences from the English Wikipedia, which were distorted using a realistic error model. Measuring the quality of an algorithm with respect to these error categories requires an alignment of the original text, the distorted text and the corrected text provided by the tool. We make our benchmark generation and evaluation tools publicly available."
https://www.aclweb.org/anthology/2020.lrec-1.228.pdf
Qiantong Xu, Alexei Baevski, Tatiana Likhomanenko, Paden Tomasello, Alexis Conneau, Ronan Collobert, Gabriel Synnaeve, Michael Auli
"Self-training and unsupervised pre-training have emerged as effective approaches to improve speech recognition systems using unlabeled data. However, it is not clear whether they learn similar patterns or if they can be effectively combined. In this paper, we show that pseudo-labeling and pre-training with wav2vec 2.0 are complementary in a variety of labeled data setups. Using just 10 minutes of labeled data from Libri-light as well as 53k hours of unlabeled data from LibriVox achieves WERs of 3.0%/5.2% on the clean and other test sets of Librispeech - rivaling the best published systems trained on 960 hours of labeled data only a year ago. Training on all labeled data of Librispeech achieves WERs of 1.5%/3.1%."
https://arxiv.org/abs/2010.11430
Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli
"We show for the first time that learning powerful representations from speech audio alone followed by fine-tuning on transcribed speech can outperform the best semi-supervised methods while being conceptually simpler. wav2vec 2.0 masks the speech input in the latent space and solves a contrastive task defined over a quantization of the latent representations which are jointly learned. Experiments using all labeled data of Librispeech achieve 1.8/3.3 WER on the clean/other test sets. When lowering the amount of labeled data to one hour, wav2vec 2.0 outperforms the previous state of the art on the 100 hour subset while using 100 times less labeled data. Using just ten minutes of labeled data and pre-training on 53k hours of unlabeled data still achieves 4.8/8.2 WER. This demonstrates the feasibility of speech recognition with limited amounts of labeled data."
https://arxiv.org/abs/2006.11477
Gabriel Synnaeve, Qiantong Xu, Jacob Kahn, Tatiana Likhomanenko, Edouard Grave, Vineel Pratap, Anuroop Sriram, Vitaliy Liptchinsky, Ronan Collobert
"We study pseudo-labeling for the semi-supervised training of ResNet, Time-Depth Separable ConvNets, and Transformers for speech recognition, with either CTC or Seq2Seq loss functions. We perform experiments on the standard LibriSpeech dataset, and leverage additional unlabeled data from LibriVox through pseudo-labeling. We show that while Transformer-based acoustic models have superior performance with the supervised dataset alone, semi-supervision improves all models across architectures and loss functions and bridges much of the performance gaps between them. In doing so, we reach a new state-of-the-art for end-to-end acoustic models decoded with an external language model in the standard supervised learning setting, and a new absolute state-of-the-art with semi-supervised training. Finally, we study the effect of leveraging different amounts of unlabeled audio, propose several ways of evaluating the characteristics of unlabeled audio which improve acoustic modeling, and show that acoustic models trained with more audio rely less on external language models."
https://arxiv.org/abs/1911.08460
Patrick von Platen
"Transformer-based encoder-decoder models were proposed in Vaswani et al. (2017) and have recently experienced a surge of interest, e.g. Lewis et al. (2019), Raffel et al. (2019), Zhang et al. (2020), Zaheer et al. (2020), Yan et al. (2020). Similar to BERT and GPT2, massive pre-trained encoder-decoder models have shown to significantly boost performance on a variety of sequence-to-sequence tasks Lewis et al. (2019), Raffel et al. (2019). However, due to the enormous computational cost attached to pre-training encoder-decoder models, the development of such models is mainly limited to large companies and institutes. [...]"
https://huggingface.co/blog/warm-starting-encoder-decoder
Ana L. L. M. Maia, Frank D. Julca-Aguilar and Nina S. T. Hirata
"We propose a new approach for segmenting a document image into its page components (e.g. text, graphics and tables). Our approach consists of two main steps. In the first step, a set of scores corresponding to the output of a convolutional neural network, one for each of the possible page component categories, is assigned to each connected component in the document. The labeled connected components define a fuzzy over-segmentation of the page. In the second step, spatially close connected components that are likely to belong to a same page component are grouped together. This is done by building an attributed region adjacency graph of the connected components and modeling the problem as an edge removal problem. Edges are then kept or removed based on a pre-trained classifier. The resulting groups, defined by the connected subgraphs, correspond to the detected page components. We evaluate our method on the ICDAR2009 dataset. Results show that our method effectively segments pages, being able to detect the nine types of page components. Furthermore, as our approach is based on simple machine learning models and graph-based techniques, it should be easily adapted to the segmentation of a variety of document types."
Keita Kurita, Nidhi Vyas, Ayush Pareek, Alan W Black, Yulia Tsvetkov
"Contextual word embeddings such as BERT have achieved state of the art performance in numerous NLP tasks. Since they are optimized to capture the statistical properties of training data, they tend to pick up on and amplify social stereotypes present in the data as well. In this study, we (1) propose a template-based method to quantify bias in BERT; (2) show that this method obtains more consistent results in capturing social biases than the traditional cosine based method; and (3) conduct a case study, evaluating gender bias in a downstream task of Gender Pronoun Resolution. Although our case study focuses on gender bias, the proposed technique is generalizable to unveiling other biases, including in multiclass settings, such as racial and religious biases."
https://www.aclweb.org/anthology/W19-3823/
Vineel Pratap, Awni Hannun, Qiantong Xu, Jeff Cai, Jacob Kahn, Gabriel Synnaeve, Vitaliy Liptchinsky, Ronan Collobert
"This paper introduces wav2letter++, the fastest open-source deep learning speech recognition framework. wav2letter++ is written entirely in C++, and uses the ArrayFire tensor library for maximum efficiency. Here we explain the architecture and design of the wav2letter++ system and compare it to other major open-source speech recognition systems. In some cases wav2letter++ is more than 2x faster than other optimized frameworks for training end-to-end neural networks for speech recognition. We also show that wav2letter++'s training times scale linearly to 64 GPUs, the highest we tested, for models with 100 million parameters. High-performance frameworks enable fast iteration, which is often a crucial factor in successful research and model tuning on new datasets and tasks."
https://arxiv.org/abs/1812.07625
Yichen Wu, Ayfer Calis, Yi Luo, Cheng Chen, Maxwell Lutton, Yair Rivenson, Xing Lin, Hatice Ceylan Koydemir, Yibo Zhang, Hongda Wang, Zoltán Göröcs, Aydogan Ozcan
"Conventional bioaerosol sensing requires the sampled aerosols in the field to be transferred to a laboratory for manual inspection, which can be rather costly and slow, also requiring a professional for labeling and microscopic examination of the samples. Here we demonstrate label-free bioaerosol sensing using a field-portable and cost-effective device based on holographic microscopy and deep-learning, which screens bioaerosols at a throughput of 13 L/min. Two different deep neural networks are designed to rapidly reconstruct the amplitude and phase images of the captured bioaerosols, and to classify the type of each bioaerosol that is imaged. As a proof-of-concept, we studied label-free sensing of common bioaerosol types, for example, Bermuda grass pollen, oak tree pollen, ragweed pollen, Aspergillus spore, and Alternaria spore and achieved >94% classification accuracy. The presented label-free bioaerosol measurement device, with its mobility and cost-effectiveness, will find several applications in indoor and outdoor air quality monitoring."
https://pubs.acs.org/doi/pdf/10.1021/acsphotonics.8b01109
Petar Ristoski, Jessica Rosati, Tommaso Di Noia, Renato De Leone, Heiko Paulheim
"Linked Open Data has been recognized as a valuable source for background information in many data mining andinformation retrieval tasks. However, most of the existing tools require features in propositional form, i.e., a vector of nominal ornumerical features associated with an instance, while Linked Open Data sources are graphs by nature. In this paper, we presentRDF2Vec, an approach that uses language modeling approaches for unsupervised feature extraction from sequences of words,and adapts them to RDF graphs. We generate sequences by leveraging local information from graph sub-structures, harvested byWeisfeiler-Lehman Subtree RDF Graph Kernels and graph walks, and learn latent numerical representations of entities in RDFgraphs. We evaluate our approach on three different tasks: (i) standard machine-learning tasks (ii) entity and document modeling(iii) content-based recommender systems. The evaluation shows that the proposed entity embeddings outperform existing tech-niques, and that feature vector representations of general knowledge graphs such as DBpedia and Wikidata can be easily reusedfor different tasks."
http://www.semantic-web-journal.net/system/files/swj1495.pdf
Wikipedia2Vec: An Efficient Toolkit for Learning and Visualizing the Embeddings of Words and Entities from Wikipedia
Ikuya Yamada, Akari Asai, Jin Sakuma, Hiroyuki Shindo, Hideaki Takeda, Yoshiyasu Takefuji, Yuji Matsumoto
"The embeddings of entities in a large knowledge base (e.g., Wikipedia) are highly beneficial for solving various natural language tasks that involve real world knowledge. In this paper, we present Wikipedia2Vec, a Python-based open-source tool for learning the embeddings of words and entities from Wikipedia. The proposed tool enables users to learn the embeddings efficiently by issuing a single command with a Wikipedia dump file as an argument. We also introduce a web-based demonstration of our tool that allows users to visualize and explore the learned embeddings. In our experiments, our tool achieved a state-of-the-art result on the KORE entity relatedness dataset, and competitive results on various standard benchmark datasets. Furthermore, our tool has been used as a key component in various recent studies. We publicize the source code, demonstration, and the pretrained embeddings for 12 languages at this https URL."
https://arxiv.org/abs/1812.06280
Jose G. Moreno, Romaric Besancon, Romain Beaumont, Eva D’hondt, Anne-Laure Ligozat, Sophie Rosset, Xavier Tannier, and Brigitte Grau
"The correct identification of the link between an entity men-tion in a text and a known entity in a large knowledge base is importantin information retrieval or information extraction. The general approachfor this task is to generate, for a given mention, a set of candidate en-tities from the base and, in a second step, determine which is the bestone. This paper proposes a novel method for the second step which isbased on the joint learning of embeddings for the words in the text andthe entities in the knowledge base. By learning these embeddings in thesame space we arrive at a more conceptually grounded model that canbe used for candidate selection based on the surrounding context. Therelative improvement of this approach is experimentally validated on arecent benchmark corpus from the TAC-EDL 2015 evaluation campaign."
https://perso.limsi.fr/bg/fichiers/2017/combining-word-entity-eswc2017.pdf
Hamed Shahbazi, Xiaoli Z. Fern, Reza Ghaeini,Rasha Obeidat, Prasad Tadepalli
"We present a new local entity disambiguation system. The key to our system is anovel approach for learning entity represen-tations. In our approach we learn an entityaware extension of Embedding for LanguageModel (ELMo) which we call Entity-ELMo(E-ELMo). Given a paragraph containing oneor more named entity mentions, each men-tion is first defined as a function of the en-tire paragraph (including other mentions), thenthey predict the referent entities. Utilizing E-ELMo for local entity disambiguation, we out-perform all of the state-of-the-art local andglobal models on the popular benchmarks byimproving about 0.5% on micro average accuracy for AIDA test-b with Yago candidate set.The evaluation setup of the training data andcandidate set are the same as our baselines forfair comparison."
https://arxiv.org/abs/1908.05762
Samuel Broscheit
"A typical architecture for end-to-end entity linking systems consists of three steps: mention detection, candidate generation and entity disambiguation. In this study we investigate the following questions: (a) Can all those steps be learned jointly with a model for contextualized text-representations, i.e. BERT? (b) How much entity knowledge is already contained in pretrained BERT? (c) Does additional entity knowledge improve BERT’s performance in downstream tasks? To this end we propose an extreme simplification of the entity linking setup that works surprisingly well: simply cast it as a per token classification over the entire entity vocabulary (over 700K classes in our case). We show on an entity linking benchmark that (i) this model improves the entity representations over plain BERT, (ii) that it outperforms entity linking architectures that optimize the tasks separately and (iii) that it only comes second to the current state-of-the-art that does mention detection and entity disambiguation jointly. Additionally, we investigate the usefulness of entity-aware token-representations in the text-understanding benchmark GLUE, as well as the question answering benchmarks SQUAD~V2 and SWAG and also the EN-DE WMT14 machine translation benchmark. To our surprise, we find that most of those benchmarks do not benefit from additional entity knowledge, except for a task with very small training data, the RTE task in GLUE, which improves by 2%."
https://www.aclweb.org/anthology/K19-1063/
"Recently, multilingual question answering became a crucial research topic, and it is receiving increased interest in the NLP community. However, the unavailability of large-scale datasets makes it challenging to train multilingual QA systems with performance comparable to the English ones. In this work, we develop the Translate Align Retrieve (TAR) method to automatically translate the Stanford Question Answering Dataset (SQuAD) v1.1 to Spanish. We then used this dataset to train Spanish QA systems by fine-tuning a Multilingual-BERT model. Finally, we evaluated our QA models with the recently proposed MLQA and XQuAD benchmarks for cross-lingual Extractive QA. Experimental results show that our models outperform the previous Multilingual-BERT baselines achieving the new state-of-the-art value of 68.1 F1 points on the Spanish MLQA corpus and 77.6 F1 and 61.8 Exact Match points on the Spanish XQuAD corpus. The resulting, synthetically generated SQuAD-es v1.1 corpora, with almost 100% of data contained in the original English version, to the best of our knowledge, is the first large-scale QA training resource for Spanish."
"There’s something magical about Recurrent Neural Networks (RNNs). I still remember when I trained my first recurrent network for Image Captioning. Within a few dozen minutes of training my first baby model (with rather arbitrarily-chosen hyperparameters) started to generate very nice looking descriptions of images that were on the edge of making sense. Sometimes the ratio of how simple your model is to the quality of the results you get out of it blows past your expectations, and this was one of those times. What made this result so shocking at the time was that the common wisdom was that RNNs were supposed to be difficult to train (with more experience I’ve in fact reached the opposite conclusion). Fast forward about a year: I’m training RNNs all the time and I’ve witnessed their power and robustness many times, and yet their magical outputs still find ways of amusing me. This post is about sharing some of that magic with you."
http://karpathy.github.io/2015/05/21/rnn-effectiveness/
Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, Radu Soricut
"Increasing model size when pretraining natural language representations often results in improved performance on downstream tasks. However, at some point further model increases become harder due to GPU/TPU memory limitations, longer training times, and unexpected model degradation. To address these problems, we present two parameter-reduction techniques to lower memory consumption and increase the training speed of BERT. Comprehensive empirical evidence shows that our proposed methods lead to models that scale much better compared to the original BERT. We also use a self-supervised loss that focuses on modeling inter-sentence coherence, and show it consistently helps downstream tasks with multi-sentence inputs. As a result, our best model establishes new state-of-the-art results on the GLUE, RACE, and SQuAD benchmarks while having fewer parameters compared to BERT-large."
https://arxiv.org/abs/1909.11942
"Segmenting newspaper pages into articles that semantically belong together is a necessary prerequisite for article-based information retrieval on print media collections like e.g. archives and libraries. It is challenging due to vastly differing layouts of papers, various content types and different languages, but commercially very relevant for e.g. media monitoring. We present a semantic segmentation approach based on the visual appearance of each page. We apply a fully convolutional neural network (FCN) that we train in an end-to-end fashion to transform the input image into a segmentation mask in one pass. We show experimentally that the FCN performs very well: it outperforms a deep learning-based commercial solution by a large margin in terms of segmentation quality while in addition being computationally two orders of magnitude more efficient."
https://pd.zhaw.ch/publikation/upload/212962.pdf
"The dominant sequence transduction models are based on complex recurrent or convolutional neural networks in an encoder-decoder configuration. The best performing models also connect the encoder and decoder through an attention mechanism. We propose a new simple network architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence and convolutions entirely. Experiments on two machine translation tasks show these models to be superior in quality while being more parallelizable and requiring significantly less time to train. Our model achieves 28.4 BLEU on the WMT 2014 English-to-German translation task, improving over the existing best results, including ensembles by over 2 BLEU. On the WMT 2014 English-to-French translation task, our model establishes a new single-model state-of-the-art BLEU score of 41.8 after training for 3.5 days on eight GPUs, a small fraction of the training costs of the best models from the literature. We show that the Transformer generalizes well to other tasks by applying it successfully to English constituency parsing both with large and limited training data."
https://arxiv.org/abs/1706.03762
"Generating long pieces of music is a challenging problem, as music contains structure at multiple timescales, from milisecond timings to motifs to phrases to repetition of entire sections. We present Music Transformer, an attention-based neural network that can generate music with improved long-term coherence"
https://magenta.tensorflow.org/music-transformer
"Language model pre-training has proven to be useful in learning universal language representations. As a state-of-the-art language model pre-training model, BERT (Bidirectional Encoder Representations from Transformers) has achieved amazing results in many language understanding tasks. In this paper, we conduct exhaustive experiments to investigate different fine-tuning methods of BERT on text classification task and provide a general solution for BERT fine-tuning. Finally, the proposed solution obtains new state-of-the-art results on eight widely-studied text classification datasets."
"In this paper we propose a new method for automated segmentation of scanned newspaper pages into articles. Article regions are produced as a result of merging sub-article level content and title regions. We use a Bayesian Gaussian mixture model to model page Connected Component information and cluster input into sub-article components. The Bayesian model is conditioned on a prior distribution over region features, aiding classification into titles and content. Using a Dirichlet prior we are able to automatically estimate correctly the number of title and article regions. The method is tested on a dataset of digitized historical newspapers, where visual experimental results are very promising."
https://dl.acm.org/citation.cfm?doid=2960811.2967165
"Semi-supervised learning lately has shown much promise in improving deep learning models when labeled data is scarce. Common among recent approaches is the use of consistency training on a large amount of unlabeled data to constrain model predictions to be invariant to input noise. In this work, we present a new perspective on how to effectively noise unlabeled examples and argue that the quality of noising, specifically those produced by advanced data augmentation methods, plays a crucial role in semi-supervised learning. By substituting simple noising operations with advanced data augmentation methods, our method brings substantial improvements across six language and three vision tasks under the same consistency training framework. On the IMDb text classification dataset, with only 20 labeled examples, our method achieves an error rate of 4.20, outperforming the state-of-the-art model trained on 25,000 labeled examples. On a standard semi-supervised learning benchmark, CIFAR-10, our method outperforms all previous approaches and achieves an error rate of 2.7% with only 4,000 examples, nearly matching the performance of models trained on 50,000 labeled examples. Our method also combines well with transfer learning, e.g., when finetuning from BERT, and yields improvements in high-data regime, such as ImageNet, whether when there is only 10% labeled data or when a full labeled set with 1.3M extra unlabeled examples is used."
https://arxiv.org/abs/1904.12848
"BERT became an essential ingredient of many NLP deep learning pipelines. It is considered a milestone in NLP, as ResNet is in the computer vision field. The only problem with BERT is its size. BERT-base is model contains 110M parameters. The larger variant BERT-large contains 340M parameters. It’s hard to deploy a model of such size into many environments with limited resources, such as a mobile or embedded systems."
https://blog.inten.to/speeding-up-bert-5528e18bb4ea
"As Transfer Learning from large-scale pre-trained models becomes more prevalent in Natural Language Processing (NLP), operating these large models in on-the-edge and/or under constrained computational training or inference budgets remains challenging. In this work, we propose a method to pre-train a smaller general-purpose language representation model, called DistilBERT, which can then be fine-tuned with good performances on a wide range of tasks like its larger counterparts. While most prior work investigated the use of distillation for building task-specific models, we leverage knowledge distillation during the pre-training phase and show that it is possible to reduce the size of a BERT model by 40%, while retaining 97% of its language understanding capabilities and being 60% faster. To leverage the inductive biases learned by larger models during pre-training, we introduce a triple loss combining language modeling, distillation and cosine-distance losses. Our smaller, faster and lighter model is cheaper to pre-train and we demonstrate its capabilities for on-device computations in a proof-of-concept experiment and a comparative on-device study."
https://arxiv.org/abs/1910.01108
"Language model pre-training, such as BERT, has significantly improved the performances of many natural language processing tasks. However, pre-trained language models are usually computationally expensive and memory intensive, so it is difficult to effectively execute them on some resource-restricted devices. To accelerate inference and reduce model size while maintaining accuracy, we firstly propose a novel transformer distillation method that is a specially designed knowledge distillation (KD) method for transformer-based models. By leveraging this new KD method, the plenty of knowledge encoded in a large teacher BERT can be well transferred to a small student TinyBERT. Moreover, we introduce a new two-stage learning framework for TinyBERT, which performs transformer distillation at both the pre-training and task-specific learning stages. This framework ensures that TinyBERT can capture both the general-domain and task-specific knowledge of the teacher BERT.TinyBERT is empirically effective and achieves more than 96% the performance of teacher BERTBASE on GLUE benchmark while being 7.5x smaller and 9.4x faster on inference. TinyBERT is also significantly better than state-of-the-art baselines on BERT distillation, with only about 28% parameters and about 31% inference time of them."
https://arxiv.org/abs/1909.10351
Ashok Chilakapati
"Word vectors have evolved over the years to know the difference between “record the play” vs “play the record”. They have evolved from a one-hot world where every word was orthogonal to every other word, to a place where word vectors morph to suit the context. Slapping a BoW on word vectors is the usual way to build a document vector for tasks such as classification. But BERT does not need a BoW as the vector shooting out of the top [CLS] token is already primed for the specific classification objective ..."
https://xplordat.com/2019/09/23/bow-to-bert/
https://medium.com/@adriensieg/text-similarities-da019229c894