An emerging direction of quantum computing is to establish meaningful quantum applications in various fields of artificial intelligence, including natural language processing (NLP). Although some efforts based on syntactic analysis have opened the door to research in Quantum NLP (QNLP), limitations such as heavy syntactic preprocessing and syntax-dependent network architecture make them impracticable on larger and real-world data sets. In this paper, we propose a new simple network architecture, called the quantum self-attention neural network (QSANN), which can make up for these limitations. Specifically, we introduce the self-attention mechanism into quantum neural networks and then utilize a Gaussian projected quantum self-attention serving as a sensible quantum version of self-attention. As a result, QSANN is effective and scalable on larger data sets and has the desirable property of being implementable on near-term quantum devices. In particular, our QSANN outperforms the best existing QNLP model based on syntactic analysis as well as a simple classical self-attention neural network in numerical experiments of text classification tasks on public data sets. We further show that our method exhibits robustness to low-level quantum noises.
Handwritten text recognition has been widely studied in the last decades for its numerous applications. Nowadays, the state-of-the-art approach consists in a three-step process. The document is segmented into text lines, which are then ordered and recognized. However, this three-step approach has many drawbacks. The three steps are treated independently whereas they are closely related. Errors accumulate from one step to the other. The ordering step is based on heuristic rules which prevent its use for documents with a complex layouts or for heterogeneous documents. The need for additional physical segmentation annotations for training the segmentation stage is inherent to this approach. In this thesis, we propose to tackle these issues by performing the handwritten text recognition of whole document in an end-to-end way. To this aim, we gradually increase the difficulty of the recognition task, moving from isolated lines to paragraphs, and then to whole documents. We proposed an approach at the line level, based on a fully convolutional network, in order to design a first generic feature extraction step for the handwriting recognition task. Based on this preliminary work, we studied two different approaches to recognize handwritten paragraphs. We reached state-of-the-art results at paragraph level on the RIMES 2011, IAM and READ 2016 datasets and outperformed the line-level state of the art on these datasets. We finally proposed the first end-to-end approach dedicated to the recognition of both text and layout, at document level. Characters and layout tokens are sequentially predicted following a learned reading order. We proposed two new metrics we used to evaluate this task on the RIMES 2009 and READ 2016 dataset, at page level and double-page level.
Text tracking is to track multiple texts in a video,and construct a trajectory for each text. Existing methodstackle this task by utilizing the tracking-by-detection frame-work, i.e., detecting the text instances in each frame andassociating the corresponding text instances in consecutiveframes. We argue that the tracking accuracy of this paradigmis severely limited in more complex scenarios, e.g., owing tomotion blur, etc., the missed detection of text instances causesthe break of the text trajectory. In addition, different textinstances with similar appearance are easily confused, leadingto the incorrect association of the text instances. To this end,a novel spatio-temporal complementary text tracking model isproposed in this paper. We leverage a Siamese ComplementaryModule to fully exploit the continuity characteristic of the textinstances in the temporal dimension, which effectively alleviatesthe missed detection of the text instances, and hence ensuresthe completeness of each text trajectory. We further integratethe semantic cues and the visual cues of the text instance intoa unified representation via a text similarity learning network,which supplies a high discriminative power in the presence oftext instances with similar appearance, and thus avoids the mis-association between them. Our method achieves state-of-the-art performance on several public benchmarks. The source codeis available at https://github.com/lsabrinax/VideoTextSCM.
Video-Text pre-training aims at learning transferable representations from large-scale video-text pairs via aligning the semantics between visual and textual information. State-of-the-art approaches extract visual features from raw pixels in an end-to-end fashion. However, these methods operate at frame-level directly and thus overlook the spatio-temporal structure of objects in video, which yet has a strong synergy with nouns in textual descriptions. In this work, we propose a simple yet effective module for video-text representation learning, namely RegionLearner, which can take into account the structure of objects during pre-training on large-scale video-text pairs. Given a video, our module (1) first quantizes visual features into semantic clusters, then (2) generates learnable masks and uses them to aggregate the features belonging to the same semantic region, and finally (3) models the interactions between different aggregated regions. In contrast to using off-the-shelf object detectors, our proposed module does not require explicit supervision and is much more computationally efficient. We pre-train the proposed approach on the public WebVid2M and CC3M datasets. Extensive evaluations on four downstream video-text retrieval benchmarks clearly demonstrate the effectiveness of our RegionLearner. The code will be available at https://github.com/ruiyan1995/Region_Learner.
Recent studies show that auto-encoder based approaches successfully perform language generation, smooth sentence interpolation, and style transfer over unseen attributes using unlabelled datasets in a zero-shot manner. The latent space geometry of such models is organised well enough to perform on datasets where the style is "coarse-grained" i.e. a small fraction of words alone in a sentence are enough to determine the overall style label. A recent study uses a discrete token-based perturbation approach to map "similar" sentences ("similar" defined by low Levenshtein distance/ high word overlap) close by in latent space. This definition of "similarity" does not look into the underlying nuances of the constituent words while mapping latent space neighbourhoods and therefore fails to recognise sentences with different style-based semantics while mapping latent neighbourhoods. We introduce EPAAEs (Embedding Perturbed Adversarial AutoEncoders) which completes this perturbation model, by adding a finely adjustable noise component on the continuous embeddings space. We empirically show that this (a) produces a better organised latent space that clusters stylistically similar sentences together, (b) performs best on a diverse set of text style transfer tasks than similar denoising-inspired baselines, and (c) is capable of fine-grained control of Style Transfer strength. We also extend the text style transfer tasks to NLI datasets and show that these more complex definitions of style are learned best by EPAAE. To the best of our knowledge, extending style transfer to NLI tasks has not been explored before.
A new line of research for feature selection based on neural networks has recently emerged. Despite its superiority to classical methods, it requires many training iterations to converge and detect informative features. The computational time becomes prohibitively long for datasets with a large number of samples or a very high dimensional feature space. In this paper, we present a new efficient unsupervised method for feature selection based on sparse autoencoders. In particular, we propose a new sparse training algorithm that optimizes a model's sparse topology during training to pay attention to informative features quickly. The attention-based adaptation of the sparse topology enables fast detection of informative features after a few training iterations. We performed extensive experiments on 10 datasets of different types, including image, speech, text, artificial, and biological. They cover a wide range of characteristics, such as low and high-dimensional feature spaces, and few and large training samples. Our proposed approach outperforms the state-of-the-art methods in terms of selecting informative features while reducing training iterations and computational costs substantially. Moreover, the experiments show the robustness of our method in extremely noisy environments.
One of the major challenges in training text-to-image generation models is the need of a large number of high-quality image-text pairs. While image samples are often easily accessible, the associated text descriptions typically require careful human captioning, which is particularly time- and cost-consuming. In this paper, we propose the first work to train text-to-image generation models without any text data. Our method leverages the well-aligned multi-modal semantic space of the powerful pre-trained CLIP model: the requirement of text-conditioning is seamlessly alleviated via generating text features from image features. Extensive experiments are conducted to illustrate the effectiveness of the proposed method. We obtain state-of-the-art results in the standard text-to-image generation tasks. Importantly, the proposed language-free model outperforms most existing models trained with full image-text pairs. Furthermore, our method can be applied in fine-tuning pre-trained models, which saves both training time and cost in training text-to-image generation models. Our pre-trained model obtains competitive results in zero-shot text-to-image generation on the MS-COCO dataset, yet with around only 1% of the model size and training data size relative to the recently proposed large DALL-E model.
Multimodal semantic understanding often has to deal with uncertainty, which means the obtained message tends to refer to multiple targets. Such uncertainty is problematic for our interpretation, including intra-modal and inter-modal uncertainty. Little effort studies the modeling of this uncertainty, particularly in pre-training on unlabeled datasets and fine-tuning in task-specific downstream tasks. To address this, we project the representations of all modalities as probabilistic distributions via a Probability Distribution Encoder (PDE) by utilizing rich multimodal semantic information. Furthermore, we integrate uncertainty modeling with popular pre-training frameworks and propose suitable pre-training tasks: Distribution-based Vision-Language Contrastive learning (D-VLC), Distribution-based Masked Language Modeling (D-MLM), and Distribution-based Image-Text Matching (D-ITM). The fine-tuned models are applied to challenging downstream tasks, including image-text retrieval, visual question answering, visual reasoning, and visual entailment, and achieve state-of-the-art results. Code is released at https://github.com/IIGROUP/MAP.
This paper presents work on novel machine translation (MT) systems between spoken and signed languages, where signed languages are represented in SignWriting, a sign language writing system. Our work seeks to address the lack of out-of-the-box support for signed languages in current MT systems and is based on the SignBank dataset, which contains pairs of spoken language text and SignWriting content. We introduce novel methods to parse, factorize, decode, and evaluate SignWriting, leveraging ideas from neural factored MT. In a bilingual setup--translating from American Sign Language to (American) English--our method achieves over 30 BLEU, while in two multilingual setups--translating in both directions between spoken languages and signed languages--we achieve over 20 BLEU. We find that common MT techniques used to improve spoken language translation similarly affect the performance of sign language translation. These findings validate our use of an intermediate text representation for signed languages to include them in natural language processing research.
As deep learning is now used in many real-world applications, research has focused increasingly on the privacy of deep learning models and how to prevent attackers from obtaining sensitive information about the training data. However, image-text models like CLIP have not yet been looked at in the context of privacy attacks. While membership inference attacks aim to tell whether a specific data point was used for training, we introduce a new type of privacy attack, named identity inference attack (IDIA), designed for multi-modal image-text models like CLIP. Using IDIAs, an attacker can reveal whether a particular person, was part of the training data by querying the model in a black-box fashion with different images of the same person. Letting the model choose from a wide variety of possible text labels, the attacker can probe the model whether it recognizes the person and, therefore, was used for training. Through several experiments on CLIP, we show that the attacker can identify individuals used for training with very high accuracy and that the model learns to connect the names with the depicted people. Our experiments show that a multi-modal image-text model indeed leaks sensitive information about its training data and, therefore, should be handled with care.