In the field of text-independent speaker recognition, dynamic models that change along the time axis have been proposed to consider the phoneme-varying characteristics of speech. However, detailed analysis on how dynamic models work depending on phonemes is insufficient. In this paper, we propose temporal dynamic CNN (TDY-CNN) that considers temporal variation of phonemes by applying kernels optimally adapt to each time bin. These kernels adapt to time bins by applying weighted sum of trained basis kernels. Then, an analysis on how adaptive kernels work on different phonemes in various layers is carried out. TDY-ResNet-38(x0.5) using six basis kernels shows better speaker verification performance than baseline model ResNet-38(x0.5) does, with an equal error rate (EER) of 1.48%. In addition, we showed that adaptive kernels depend on phoneme groups and more phoneme-specific at early layer. Temporal dynamic model adapts itself to phonemes without explicitly given phoneme information during training, and the results show that the necessity to consider phoneme variation within utterances for more accurate and robust text-independent speaker verification.
With Internet users constantly leaving a trail of text, whether through blogs, emails, or social media posts, the ability to write and protest anonymously is being eroded because artificial intelligence, when given a sample of previous work, can match text with its author out of hundreds of possible candidates. Existing approaches to authorship anonymization, also known as authorship obfuscation, often focus on protecting binary demographic attributes rather than identity as a whole. Even those that do focus on obfuscating identity require manual feedback, lose the coherence of the original sentence, or only perform well given a limited subset of authors. In this paper, we develop a new approach to authorship anonymization by constructing a generative adversarial network that protects identity and optimizes for three different losses corresponding to anonymity, fluency, and content preservation. Our fully automatic method achieves comparable results to other methods in terms of content preservation and fluency, but greatly outperforms baselines in regards to anonymization. Moreover, our approach is able to generalize well to an open-set context and anonymize sentences from authors it has not encountered before.
A crucial component for the scene text based reasoning required for TextVQA and TextCaps datasets involve detecting and recognizing text present in the images using an optical character recognition (OCR) system. The current systems are crippled by the unavailability of ground truth text annotations for these datasets as well as lack of scene text detection and recognition datasets on real images disallowing the progress in the field of OCR and evaluation of scene text based reasoning in isolation from OCR systems. In this work, we propose TextOCR, an arbitrary-shaped scene text detection and recognition with 900k annotated words collected on real images from TextVQA dataset. We show that current state-of-the-art text-recognition (OCR) models fail to perform well on TextOCR and that training on TextOCR helps achieve state-of-the-art performance on multiple other OCR datasets as well. We use a TextOCR trained OCR model to create PixelM4C model which can do scene text based reasoning on an image in an end-to-end fashion, allowing us to revisit several design choices to achieve new state-of-the-art performance on TextVQA dataset.
Documents often contain complex physical structures, which make the Document Layout Analysis (DLA) task challenging. As a pre-processing step for content extraction, DLA has the potential to capture rich information in historical or scientific documents on a large scale. Although many deep-learning-based methods from computer vision have already achieved excellent performance in detecting \emph{Figure} from documents, they are still unsatisfactory in recognizing the \emph{List}, \emph{Table}, \emph{Text} and \emph{Title} category blocks in DLA. This paper proposes a VTLayout model fusing the documents' deep visual, shallow visual, and text features to localize and identify different category blocks. The model mainly includes two stages, and the three feature extractors are built in the second stage. In the first stage, the Cascade Mask R-CNN model is applied directly to localize all category blocks of the documents. In the second stage, the deep visual, shallow visual, and text features are extracted for fusion to identify the category blocks of documents. As a result, we strengthen the classification power of different category blocks based on the existing localization technique. The experimental results show that the identification capability of the VTLayout is superior to the most advanced method of DLA based on the PubLayNet dataset, and the F1 score is as high as 0.9599.
The task of organizing a shuffled set of sentences into a coherent text is important in NLP and has been used to evaluate a machine's understanding of causal and temporal relations. We present Reorder-BART (RE-BART), a sentence ordering framework which leverages a pre-trained transformer-based model to identify a coherent order for a given set of shuffled sentences. We reformulate the task as a conditional text-to-marker generation setup where the input is a set of shuffled sentences with sentence-specific markers and output is a sequence of position markers of the ordered text. Our framework achieves the state-of-the-art performance across six datasets in Perfect Match Ratio (PMR) and Kendall's tau ($\tau$) metric. We perform evaluations in a zero-shot setting, showcasing that our model is able to generalize well across other datasets. We additionally perform a series of experiments to understand the functioning and explore the limitations of our framework.
Modern language models can generate high-quality short texts. However, they often meander or are incoherent when generating longer texts. These issues arise from the next-token-only language modeling objective. To address these issues, we introduce Time Control (TC), a language model that implicitly plans via a latent stochastic process. TC does this by learning a representation which maps the dynamics of how text changes in a document to the dynamics of a stochastic process of interest. Using this representation, the language model can generate text by first implicitly generating a document plan via a stochastic process, and then generating text that is consistent with this latent plan. Compared to domain-specific methods and fine-tuning GPT2 across a variety of text domains, TC improves performance on text infilling and discourse coherence. On long text generation settings, TC preserves the text structure both in terms of ordering (up to +40% better) and text length consistency (up to +17% better). Human evaluators also prefer TC's output 28.6% more than the baselines.
Inspired by progress in large-scale language modeling, we apply a similar approach towards building a single generalist agent beyond the realm of text outputs. The agent, which we refer to as Gato, works as a multi-modal, multi-task, multi-embodiment generalist policy. The same network with the same weights can play Atari, caption images, chat, stack blocks with a real robot arm and much more, deciding based on its context whether to output text, joint torques, button presses, or other tokens. In this report we describe the model and the data, and document the current capabilities of Gato.
The Strong Lottery Ticket Hypothesis (SLTH) stipulates the existence of a subnetwork within a sufficiently overparameterized (dense) neural network that -- when initialized randomly and without any training -- achieves the accuracy of a fully trained target network. Recent work by \citet{da2022proving} demonstrates that the SLTH can also be extended to translation equivariant networks -- i.e. CNNs -- with the same level of overparametrization as needed for SLTs in dense networks. However, modern neural networks are capable of incorporating more than just translation symmetry, and developing general equivariant architectures such as rotation and permutation has been a powerful design principle. In this paper, we generalize the SLTH to functions that preserve the action of the group $G$ -- i.e. $G$-equivariant network -- and prove, with high probability, that one can prune a randomly initialized overparametrized $G$-equivariant network to a $G$-equivariant subnetwork that approximates another fully trained $G$-equivariant network of fixed width and depth. We further prove that our prescribed overparametrization scheme is also optimal as a function of the error tolerance. We develop our theory for a large range of groups, including important ones such as subgroups of the Euclidean group $\text{E}(n)$ and subgroups of the symmetric group $G \leq \mathcal{S}_n$ -- allowing us to find SLTs for MLPs, CNNs, $\text{E}(2)$-steerable CNNs, and permutation equivariant networks as specific instantiations of our unified framework which completely extends prior work. Empirically, we verify our theory by pruning overparametrized $\text{E}(2)$-steerable CNNs and message passing GNNs to match the performance of trained target networks within a given error tolerance.
Unplanned intensive care unit (ICU) readmission rate is an important metric for evaluating the quality of hospital care. Efficient and accurate prediction of ICU readmission risk can not only help prevent patients from inappropriate discharge and potential dangers, but also reduce associated costs of healthcare. In this paper, we propose a new method that uses medical text of Electronic Health Records (EHRs) for prediction, which provides an alternative perspective to previous studies that heavily depend on numerical and time-series features of patients. More specifically, we extract discharge summaries of patients from their EHRs, and represent them with multiview graphs enhanced by an external knowledge graph. Graph convolutional networks are then used for representation learning. Experimental results prove the effectiveness of our method, yielding state-of-the-art performance for this task.
Aspect-based sentiment analysis (ABSA) has become a prevalent task in recent years. However, the absence of a unified framework in the present ABSA research makes it challenging to compare different models' performance fairly. Therefore, we created an open-source ABSA framework, namely PYABSA. Besides, previous efforts usually neglect the precursor aspect term extraction (ASC) subtask and focus on the aspect sentiment classification (ATE) subtask. Compared to previous works, PYABSA includes the features of aspect term extraction, aspect sentiment classification, and text classification, while multiple ABSA subtasks can be adapted to PYABSA owing to its modular architecture. To facilitate ABSA applications, PYABSAseamless integrates multilingual modelling, automated dataset annotation, etc., which are helpful in deploying ABSA services. In ASC and ATE, PYABSA provides up to 33 and 7 built-in models, respectively, while all the models provide quick training and instant inference. Besides, PYABSA contains 180K+ ABSA instances from 21 augmented ABSA datasets for applications and studies. PyABSA is available at https://github.com/yangheng95/PyABSA