Generative artificial intelligence (AI) is a widely popular technology that will have a profound impact on society and individuals. Less than a decade ago, it was thought that creative work would be among the last to be automated - yet today, we see AI encroaching on many creative domains. In this paper, we present the findings of a survey study on people's perceptions of text-to-image generation. We touch on participants' technical understanding of the emerging technology, their fears and concerns, and thoughts about risks and dangers of text-to-image generation to the individual and society. We find that while participants were aware of the risks and dangers associated with the technology, only few participants considered the technology to be a personal risk. The risks for others were more easy to recognize for participants. Artists were particularly seen at risk. Participants who had tried the technology rated its future importance lower than those who had not tried it. This result shows that many people are still oblivious of the potential personal risks of generative artificial intelligence and the impending societal changes associated with this technology.
Although psycholinguists and psychologists have long studied the tendency of linguistic strings to evoke mental images in hearers or readers, most computational studies have applied this concept of imageability only to isolated words. Using recent developments in text-to-image generation models, such as DALLE mini, we propose computational methods that use generated images to measure the imageability of both single English words and connected text. We sample text prompts for image generation from three corpora: human-generated image captions, news article sentences, and poem lines. We subject these prompts to different deformances to examine the model's ability to detect changes in imageability caused by compositional change. We find high correlation between the proposed computational measures of imageability and human judgments of individual words. We also find the proposed measures more consistently respond to changes in compositionality than baseline approaches. We discuss possible effects of model training and implications for the study of compositionality in text-to-image models.
Most existing text generation models follow the sequence-to-sequence paradigm. Generative Grammar suggests that humans generate natural language texts by learning language grammar. We propose a syntax-guided generation schema, which generates the sequence guided by a constituency parse tree in a top-down direction. The decoding process can be decomposed into two parts: (1) predicting the infilling texts for each constituent in the lexicalized syntax context given the source sentence; (2) mapping and expanding each constituent to construct the next-level syntax context. Accordingly, we propose a structural beam search method to find possible syntax structures hierarchically. Experiments on paraphrase generation and machine translation show that the proposed method outperforms autoregressive baselines, while also demonstrating effectiveness in terms of interpretability, controllability, and diversity.
Co-occurrence networks are an important method in the field of natural language processing and text mining for discovering semantic relationships within texts. However, the traditional traversal algorithm for constructing co-occurrence networks has high time complexity and space complexity when dealing with large-scale text data. In this paper, we propose an optimized algorithm based on inverted indexing and breadth-first search to improve the efficiency of co-occurrence network construction and reduce memory consumption. Firstly, the traditional traversal algorithm is analyzed, and its performance issues in constructing co-occurrence networks are identified. Then, the detailed implementation process of the optimized algorithm is presented. Subsequently, the CSL large-scale Chinese scientific literature dataset is used for experimental validation, comparing the performance of the traditional traversal algorithm and the optimized algorithm in terms of running time and memory usage. Finally, using non-parametric test methods, the optimized algorithm is proven to have significantly better performance than the traditional traversal algorithm. The research in this paper provides an effective method for the rapid construction of co-occurrence networks, contributing to the further development of the Information Organization fields.
Although audio generation shares commonalities across different types of audio, such as speech, music, and sound effects, designing models for each type requires careful consideration of specific objectives and biases that can significantly differ from those of other types. To bring us closer to a unified perspective of audio generation, this paper proposes a framework that utilizes the same learning method for speech, music, and sound effect generation. Our framework introduces a general representation of audio, called language of audio (LOA). Any audio can be translated into LOA based on AudioMAE, a self-supervised pre-trained representation learning model. In the generation process, we translate any modalities into LOA by using a GPT-2 model, and we perform self-supervised audio generation learning with a latent diffusion model conditioned on LOA. The proposed framework naturally brings advantages such as in-context learning abilities and reusable self-supervised pretrained AudioMAE and latent diffusion models. Experiments on the major benchmarks of text-to-audio, text-to-music, and text-to-speech demonstrate new state-of-the-art or competitive performance to previous approaches. Our demo and code are available at https://audioldm.github.io/audioldm2.
Food classification is an important task in health care. In this work, we propose a multimodal classification framework that uses the modified version of EfficientNet with the Mish activation function for image classification, and the traditional BERT transformer-based network is used for text classification. The proposed network and the other state-of-the-art methods are evaluated on a large open-source dataset, UPMC Food-101. The experimental results show that the proposed network outperforms the other methods, a significant difference of 11.57% and 6.34% in accuracy is observed for image and text classification, respectively, when compared with the second-best performing method. We also compared the performance in terms of accuracy, precision, and recall for text classification using both machine learning and deep learning-based models. The comparative analysis from the prediction results of both images and text demonstrated the efficiency and robustness of the proposed approach.
Vision-Language Pretraining (VLP) has demonstrated remarkable capabilities in learning visual representations from textual descriptions of images without annotations. Yet, effective VLP demands large-scale image-text pairs, a resource that suffers scarcity in the medical domain. Moreover, conventional VLP is limited to 2D images while medical images encompass diverse modalities, often in 3D, making the learning process more challenging. To address these challenges, we present Generative Text-Guided 3D Vision-Language Pretraining for Unified Medical Image Segmentation (GTGM), a framework that extends of VLP to 3D medical images without relying on paired textual descriptions. Specifically, GTGM utilizes large language models (LLM) to generate medical-style text from 3D medical images. This synthetic text is then used to supervise 3D visual representation learning. Furthermore, a negative-free contrastive learning objective strategy is introduced to cultivate consistent visual representations between augmented 3D medical image patches, which effectively mitigates the biases associated with strict positive-negative sample pairings. We evaluate GTGM on three imaging modalities - Computed Tomography (CT), Magnetic Resonance Imaging (MRI), and electron microscopy (EM) over 13 datasets. GTGM's superior performance across various medical image segmentation tasks underscores its effectiveness and versatility, by enabling VLP extension into 3D medical imagery while bypassing the need for paired text.
Vision-language pretraining models have achieved great success in supporting multimedia applications by understanding the alignments between images and text. While existing vision-language pretraining models primarily focus on understanding single image associated with a single piece of text, they often ignore the alignment at the intra-document level, consisting of multiple sentences with multiple images. In this work, we propose DocumentCLIP, a salience-aware contrastive learning framework to enforce vision-language pretraining models to comprehend the interaction between images and longer text within documents. Our model is beneficial for the real-world multimodal document understanding like news article, magazines, product descriptions, which contain linguistically and visually richer content. To the best of our knowledge, we are the first to explore multimodal intra-document links by contrastive learning. In addition, we collect a large Wikipedia dataset for pretraining, which provides various topics and structures. Experiments show DocumentCLIP not only outperforms the state-of-the-art baselines in the supervised setting, but also achieves the best zero-shot performance in the wild after human evaluation. Our code is available at https://github.com/FuxiaoLiu/DocumentCLIP.
Sign language, which conveys meaning through gestures, is the chief means of communication among deaf people. Recognizing sign language in natural settings presents significant challenges due to factors such as lighting, background clutter, and variations in signer characteristics. In this thesis, I study automatic sign language processing in the wild, using signing videos collected from the Internet. This thesis contributes new datasets, tasks, and methods. Most chapters of this thesis address tasks related to fingerspelling, an important component of sign language and yet has not been studied widely by prior work. I present three new large-scale ASL datasets in the wild: ChicagoFSWild, ChicagoFSWild+, and OpenASL. Using ChicagoFSWild and ChicagoFSWild+, I address fingerspelling recognition, which consists of transcribing fingerspelling sequences into text. I propose an end-to-end approach based on iterative attention that allows recognition from a raw video without explicit hand detection. I further show that using a Conformer-based network jointly modeling handshape and mouthing can bring performance close to that of humans. Next, I propose two tasks for building real-world fingerspelling-based applications: fingerspelling detection and search. For fingerspelling detection, I introduce a suite of evaluation metrics and a new detection model via multi-task training. To address the problem of searching for fingerspelled keywords in raw sign language videos, we propose a novel method that jointly localizes and matches fingerspelling segments to text. Finally, I will describe a benchmark for large-vocabulary open-domain sign language translation based on OpenASL. To address the challenges of sign language translation in realistic settings, we propose a set of techniques including sign search as a pretext task for pre-training and fusion of mouthing and handshape features.
Transit riders' feedback provided in ridership surveys, customer relationship management (CRM) channels, and in more recent times, through social media is key for transit agencies to better gauge the efficacy of their services and initiatives. Getting a holistic understanding of riders' experience through the feedback shared in those instruments is often challenging, mostly due to the open-ended, unstructured nature of text feedback. In this paper, we propose leveraging traditional transit CRM feedback to develop and deploy a transit-topic-aware large language model (LLM) capable of classifying open-ended text feedback to relevant transit-specific topics. First, we utilize semi-supervised learning to engineer a training dataset of 11 broad transit topics detected in a corpus of 6 years of customer feedback provided to the Washington Metropolitan Area Transit Authority (WMATA). We then use this dataset to train and thoroughly evaluate a language model based on the RoBERTa architecture. We compare our LLM, MetRoBERTa, to classical machine learning approaches utilizing keyword-based and lexicon representations. Our model outperforms those methods across all evaluation metrics, providing an average topic classification accuracy of 90%. Finally, we provide a value proposition of this work demonstrating how the language model, alongside additional text processing tools, can be applied to add structure to open-ended text sources of feedback like Twitter. The framework and results we present provide a pathway for an automated, generalizable approach for ingesting, visualizing, and reporting transit riders' feedback at scale, enabling agencies to better understand and improve customer experience.