Large-scale language models have been reducing the gap between machines and humans in understanding the real world, yet understanding an individual's theory of mind and behavior from text is far from being resolved. This research proposes a neural model -- Subjective Ground Attention -- that learns subjective grounds of individuals and accounts for their judgments on situations of others posted on social media. Using simple attention modules as well as taking one's previous activities into consideration, we empirically show that our model provides human-readable explanations of an individual's subjective preference in judging social situations. We further qualitatively evaluate the explanations generated by the model and claim that our model learns an individual's subjective orientation towards abstract moral concepts
Estimating the software projects' efforts developed by agile methods is important for project managers or technical leads. It provides a summary as a first view of how many hours and developers are required to complete the tasks. There are research works on automatic predicting the software efforts, including Term Frequency Inverse Document Frequency (TFIDF) as the traditional approach for this problem. Graph Neural Network is a new approach that has been applied in Natural Language Processing for text classification. The advantages of Graph Neural Network are based on the ability to learn information via graph data structure, which has more representations such as the relationships between words compared to approaches of vectorizing sequence of words. In this paper, we show the potential and possible challenges of Graph Neural Network text classification in story point level estimation. By the experiments, we show that the GNN Text Level Classification can achieve as high accuracy as about 80 percent for story points level classification, which is comparable to the traditional approach. We also analyze the GNN approach and point out several current disadvantages that the GNN approach can improve for this problem or other problems in software engineering.
Controlled table-to-text generation seeks to generate natural language descriptions for highlighted subparts of a table. Previous SOTA systems still employ a sequence-to-sequence generation method, which merely captures the table as a linear structure and is brittle when table layouts change. We seek to go beyond this paradigm by (1) effectively expressing the relations of content pieces in the table, and (2) making our model robust to content-invariant structural transformations. Accordingly, we propose an equivariance learning framework, which encodes tables with a structure-aware self-attention mechanism. This prunes the full self-attention structure into an order-invariant graph attention that captures the connected graph structure of cells belonging to the same row or column, and it differentiates between relevant cells and irrelevant cells from the structural perspective. Our framework also modifies the positional encoding mechanism to preserve the relative position of tokens in the same cell but enforce position invariance among different cells. Our technology is free to be plugged into existing table-to-text generation models, and has improved T5-based models to offer better performance on ToTTo and HiTab. Moreover, on a harder version of ToTTo, we preserve promising performance, while previous SOTA systems, even with transformation-based data augmentation, have seen significant performance drops. Our code is available at https://github.com/luka-group/Lattice.
Transformers have recently been shown to generate high quality images from texts. However, existing methods struggle to create high fidelity full-body images, especially multiple people. A person's pose has a high degree of freedom that is difficult to describe using words only; this creates errors in the generated image, such as incorrect body proportions and pose. We propose a pose-guided text-to-image model, using pose as an additional input constraint. Using the proposed Keypoint Pose Encoding (KPE) to encode human pose into low dimensional representation, our model can generate novel multi-person images accurately representing the pose and text descriptions provided, with minimal errors. We demonstrate that KPE is invariant to changes in the target image domain and image resolution; we show results on the Deepfashion dataset and create a new multi-person Deepfashion dataset to demonstrate the multi-capabilities of our approach.
While modern Text-to-Speech (TTS) systems can produce speech rated highly in terms of subjective evaluation, the distance between real and synthetic speech distributions remains understudied, where we use the term \textit{distribution} to mean the sample space of all possible real speech recordings from a given set of speakers; or of the synthetic samples that could be generated for the same set of speakers. We evaluate the distance of real and synthetic speech distributions along the dimensions of the acoustic environment, speaker characteristics and prosody using a range of speech processing measures and the respective Wasserstein distances of their distributions. We reduce these distribution distances along said dimensions by providing utterance-level information derived from the measures to the model and show they can be generated at inference time. The improvements to the dimensions translate to overall distribution distance reduction approximated using Automatic Speech Recognition (ASR) by evaluating the fitness of the synthetic data as training data.
For real-life applications, it is crucial that end-to-end spoken language translation models perform well on continuous audio, without relying on human-supplied segmentation. For online spoken language translation, where models need to start translating before the full utterance is spoken, most previous work has ignored the segmentation problem. In this paper, we compare various methods for improving models' robustness towards segmentation errors and different segmentation strategies in both offline and online settings and report results on translation quality, flicker and delay. Our findings on five different language pairs show that a simple fixed-window audio segmentation can perform surprisingly well given the right conditions.
An essential design decision for multilingual Neural Text-To-Speech (NTTS) systems is how to represent input linguistic features within the model. Looking at the wide variety of approaches in the literature, two main paradigms emerge, unified and separate representations. The former uses a shared set of phonetic tokens across languages, whereas the latter uses unique phonetic tokens for each language. In this paper, we conduct a comprehensive study comparing multilingual NTTS systems models trained with both representations. Our results reveal that the unified approach consistently achieves better cross-lingual synthesis with respect to both naturalness and accent. Separate representations tend to have an order of magnitude more tokens than unified ones, which may affect model capacity. For this reason, we carry out an ablation study to understand the interaction of the representation type with the size of the token embedding. We find that the difference between the two paradigms only emerges above a certain threshold embedding size. This study provides strong evidence that unified representations should be the preferred paradigm when building multilingual NTTS systems.
In this paper, we propose GlowVC: a multilingual multi-speaker flow-based model for language-independent text-free voice conversion. We build on Glow-TTS, which provides an architecture that enables use of linguistic features during training without the necessity of using them for VC inference. We consider two versions of our model: GlowVC-conditional and GlowVC-explicit. GlowVC-conditional models the distribution of mel-spectrograms with speaker-conditioned flow and disentangles the mel-spectrogram space into content- and pitch-relevant dimensions, while GlowVC-explicit models the explicit distribution with unconditioned flow and disentangles said space into content-, pitch- and speaker-relevant dimensions. We evaluate our models in terms of intelligibility, speaker similarity and naturalness for intra- and cross-lingual conversion in seen and unseen languages. GlowVC models greatly outperform AutoVC baseline in terms of intelligibility, while achieving just as high speaker similarity in intra-lingual VC, and slightly worse in the cross-lingual setting. Moreover, we demonstrate that GlowVC-explicit surpasses both GlowVC-conditional and AutoVC in terms of naturalness.
Text semantic matching is a fundamental task that has been widely used in various scenarios, such as community question answering, information retrieval, and recommendation. Most state-of-the-art matching models, e.g., BERT, directly perform text comparison by processing each word uniformly. However, a query sentence generally comprises content that calls for different levels of matching granularity. Specifically, keywords represent factual information such as action, entity, and event that should be strictly matched, while intents convey abstract concepts and ideas that can be paraphrased into various expressions. In this work, we propose a simple yet effective training strategy for text semantic matching in a divide-and-conquer manner by disentangling keywords from intents. Our approach can be easily combined with pre-trained language models (PLM) without influencing their inference efficiency, achieving stable performance improvements against a wide range of PLMs on three benchmarks.
While recent advancements in artificial intelligence (AI) language models demonstrate cutting-edge performance when working with English texts, equivalent models do not exist in other languages or do not reach the same performance level. This undesired effect of AI advancements increases the gap between access to new technology from different populations across the world. This unsought bias mainly discriminates against individuals whose English skills are less developed, e.g., non-English speakers children. Following significant advancements in AI research in recent years, OpenAI has recently presented DALL-E: a powerful tool for creating images based on English text prompts. While DALL-E is a promising tool for many applications, its decreased performance when given input in a different language, limits its audience and deepens the gap between populations. An additional limitation of the current DALL-E model is that it only allows for the creation of a few images in response to a given input prompt, rather than a series of consecutive coherent frames that tell a story or describe a process that changes over time. Here, we present an easy-to-use automatic DALL-E storytelling framework that leverages the existing DALL-E model to enable fast and coherent visualizations of non-English songs and stories, pushing the limit of the one-step-at-a-time option DALL-E currently offers. We show that our framework is able to effectively visualize stories from non-English texts and portray the changes in the plot over time. It is also able to create a narrative and maintain interpretable changes in the description across frames. Additionally, our framework offers users the ability to specify constraints on the story elements, such as a specific location or context, and to maintain a consistent style throughout the visualization.