Medical professionals frequently work in a data constrained setting to provide insights across a unique demographic. A few medical observations, for instance, informs the diagnosis and treatment of a patient. This suggests a unique setting for meta-learning, a method to learn models quickly on new tasks, to provide insights unattainable by other methods. We investigate the use of meta-learning and robustness techniques on a broad corpus of benchmark text and medical data. To do this, we developed new data pipelines, combined language models with meta-learning approaches, and extended existing meta-learning algorithms to minimize worst case loss. We find that meta-learning on text is a suitable framework for text-based data, providing better data efficiency and comparable performance to few-shot language models and can be successfully applied to medical note data. Furthermore, meta-learning models coupled with DRO can improve worst case loss across disease codes.
Scene text spotting is of great importance to the computer vision community due to its wide variety of applications. Recent methods attempt to introduce linguistic knowledge for challenging recognition rather than pure visual classification. However, how to effectively model the linguistic rules in end-to-end deep networks remains a research challenge. In this paper, we argue that the limited capacity of language models comes from 1) implicit language modeling; 2) unidirectional feature representation; and 3) language model with noise input. Correspondingly, we propose an autonomous, bidirectional and iterative ABINet++ for scene text spotting. Firstly, the autonomous suggests enforcing explicitly language modeling by decoupling the recognizer into vision model and language model and blocking gradient flow between both models. Secondly, a novel bidirectional cloze network (BCN) as the language model is proposed based on bidirectional feature representation. Thirdly, we propose an execution manner of iterative correction for the language model which can effectively alleviate the impact of noise input. Finally, to polish ABINet++ in long text recognition, we propose to aggregate horizontal features by embedding Transformer units inside a U-Net, and design a position and content attention module which integrates character order and content to attend to character features precisely. ABINet++ achieves state-of-the-art performance on both scene text recognition and scene text spotting benchmarks, which consistently demonstrates the superiority of our method in various environments especially on low-quality images. Besides, extensive experiments including in English and Chinese also prove that, a text spotter that incorporates our language modeling method can significantly improve its performance both in accuracy and speed compared with commonly used attention-based recognizers.
Current dense retrievers (DRs) are limited in their ability to effectively process misspelled queries, which constitute a significant portion of query traffic in commercial search engines. The main issue is that the pre-trained language model-based encoders used by DRs are typically trained and fine-tuned using clean, well-curated text data. Misspelled queries are typically not found in the data used for training these models, and thus misspelled queries observed at inference time are out-of-distribution compared to the data used for training and fine-tuning. Previous efforts to address this issue have focused on \textit{fine-tuning} strategies, but their effectiveness on misspelled queries remains lower than that of pipelines that employ separate state-of-the-art spell-checking components. To address this challenge, we propose ToRoDer (TypOs-aware bottlenecked pre-training for RObust DEnse Retrieval), a novel \textit{pre-training} strategy for DRs that increases their robustness to misspelled queries while preserving their effectiveness in downstream retrieval tasks. ToRoDer utilizes an encoder-decoder architecture where the encoder takes misspelled text with masked tokens as input and outputs bottlenecked information to the decoder. The decoder then takes as input the bottlenecked embeddings, along with token embeddings of the original text with the misspelled tokens masked out. The pre-training task is to recover the masked tokens for both the encoder and decoder. Our extensive experimental results and detailed ablation studies show that DRs pre-trained with ToRoDer exhibit significantly higher effectiveness on misspelled queries, sensibly closing the gap with pipelines that use a separate, complex spell-checker component, while retaining their effectiveness on correctly spelled queries.
Recent breakthroughs in the field of language-guided image generation have yielded impressive achievements, enabling the creation of high-quality and diverse images based on user instructions. Although the synthesis performance is fascinating, one significant limitation of current image generation models is their insufficient ability to generate coherent text within images, particularly for complex glyph structures like Chinese characters. To address this problem, we introduce GlyphDraw, a general learning framework aiming at endowing image generation models with the capacity to generate images embedded with coherent text. To the best of our knowledge, this is the first work in the field of image synthesis to address the generation of Chinese characters. % we first adopt the OCR technique to collect images with Chinese characters as training samples, and extract the text and locations as auxiliary information. We first sophisticatedly design the image-text dataset's construction strategy, then build our model specifically on a diffusion-based image generator and carefully modify the network structure to allow the model to learn drawing Chinese characters with the help of glyph and position information. Furthermore, we maintain the model's open-domain image synthesis capability by preventing catastrophic forgetting by using a variety of training techniques. Extensive qualitative and quantitative experiments demonstrate that our method not only produces accurate Chinese characters as in prompts, but also naturally blends the generated text into the background. Please refer to https://1073521013.github.io/glyph-draw.github.io
Advancements in Text-to-Image synthesis over recent years have focused more on improving the quality of generated samples on datasets with descriptive captions. However, real-world image-caption pairs present in domains such as news data do not use simple and directly descriptive captions. With captions containing information on both the image content and underlying contextual cues, they become abstractive in nature. In this paper, we launch ANNA, an Abstractive News captioNs dAtaset extracted from online news articles in a variety of different contexts. We explore the capabilities of current Text-to-Image synthesis models to generate news domain-specific images using abstractive captions by benchmarking them on ANNA, in both standard training and transfer learning settings. The generated images are judged on the basis of contextual relevance, visual quality, and perceptual similarity to ground-truth image-caption pairs. Through our experiments, we show that techniques such as transfer learning achieve limited success in understanding abstractive captions but still fail to consistently learn the relationships between content and context features.
Most humans use visual imagination to understand and reason about language, but models such as BERT reason about language using knowledge acquired during text-only pretraining. In this work, we investigate whether vision-and-language pretraining can improve performance on text-only tasks that involve implicit visual reasoning, focusing primarily on zero-shot probing methods. We propose a suite of visual language understanding (VLU) tasks for probing the visual reasoning abilities of text encoder models, as well as various non-visual natural language understanding (NLU) tasks for comparison. We also contribute a novel zero-shot knowledge probing method, Stroop probing, for applying models such as CLIP to text-only tasks without needing a prediction head such as the masked language modelling head of models like BERT. We show that SOTA multimodally trained text encoders outperform unimodally trained text encoders on the VLU tasks while being underperformed by them on the NLU tasks, lending new context to previously mixed results regarding the NLU capabilities of multimodal models. We conclude that exposure to images during pretraining affords inherent visual reasoning knowledge that is reflected in language-only tasks that require implicit visual reasoning. Our findings bear importance in the broader context of multimodal learning, providing principled guidelines for the choice of text encoders used in such contexts.
To align conditional text generation model outputs with desired behaviors, there has been an increasing focus on training the model using reinforcement learning (RL) with reward functions learned from human annotations. Under this framework, we identify three common cases where high rewards are incorrectly assigned to undesirable patterns: noise-induced spurious correlation, naturally occurring spurious correlation, and covariate shift. We show that even though learned metrics achieve high performance on the distribution of the data used to train the reward function, the undesirable patterns may be amplified during RL training of the text generation model. While there has been discussion about reward gaming in the RL or safety community, in this short discussion piece, we would like to highlight reward gaming in the NLG community using concrete conditional text generation examples and discuss potential fixes and areas for future work.
The impact of artificial intelligence systems on our society is increasing at an unprecedented speed. For instance, ChatGPT is being tested in mental health treatment applications such as Koko, Stable Diffusion generates pieces of art competitive with (or outperforming) human artists, and so on. Ethical concerns regarding the behavior and applications of generative AI systems have been increasing over the past years, and the field of AI alignment - steering the behavior of AI systems towards being aligned with human values - is a rapidly growing subfield of modern AI. In this paper, we address the challenges involved in ethical evaluation of a multimodal artificial intelligence system. The multimodal systems we focus on take both text and an image as input and output text, completing the sentence or answering the question asked as input. We perform the evaluation of these models in two steps: we first discus the creation of a multimodal ethical database and then use this database to construct morality-evaluating algorithms. The creation of the multimodal ethical database is done interactively through human feedback. Users are presented with multiple examples and votes on whether they are ethical or not. Once these answers have been aggregated into a dataset, we built and tested different algorithms to automatically evaluate the morality of multimodal systems. These algorithms aim to classify the answers as ethical or not. The models we tested are a RoBERTa-large classifier and a multilayer perceptron classifier.
In this study, we analyze NLG automatic metrics based on whether human evaluation aspect is used as context or objective to compute the metrics: (i) Task-agnostic and (ii) Human-aligned. Task-agnostic metrics, such as Perplexity, BLEU, BERTScore, are cost-effective and highly adaptable to diverse NLG tasks, yet they have a weak correlation with human. Human-aligned metrics (CTC, CtrlEval, UniEval) improves correlation level by incorporating desirable human-like qualities as training objective. However, their effectiveness at discerning system-level performance and quality of system outputs remains unclear. We present metric preference checklist as a framework to assess the discriminative power of automatic metrics in three NLG tasks: Text Summarization, Dialogue Response Generation, and Controlled Generation. We show that multi-aspect human-aligned metric (UniEval) is not necessarily dominant over single-aspect human-aligned metrics (CTC, CtrlEval) and task-agnostic metrics (BLEU, BERTScore), particularly when a disagreement between human evaluation aspects is present. We also show particular use cases in which automatic metrics provide a better guidance than human on discriminating system-level performance. Our proposed framework provides access: (i) for verifying whether automatic metrics are faithful to human preference, regardless their correlation level to human; and (ii) for scrutinizing the strengths and limitations of NLG systems, which are often obscured by a standard averaging method of evaluation scores.
End-to-end text image translation (TIT), which aims at translating the source language embedded in images to the target language, has attracted intensive attention in recent research. However, data sparsity limits the performance of end-to-end text image translation. Multi-task learning is a non-trivial way to alleviate this problem via exploring knowledge from complementary related tasks. In this paper, we propose a novel text translation enhanced text image translation, which trains the end-to-end model with text translation as an auxiliary task. By sharing model parameters and multi-task training, our model is able to take full advantage of easily-available large-scale text parallel corpus. Extensive experimental results show our proposed method outperforms existing end-to-end methods, and the joint multi-task learning with both text translation and recognition tasks achieves better results, proving translation and recognition auxiliary tasks are complementary.