Modern public ASR tools usually provide rich support for training various sequence-to-sequence (S2S) models, but rather simple support for decoding open-vocabulary scenarios only. For closed-vocabulary scenarios, public tools supporting lexical-constrained decoding are usually only for classical ASR, or do not support all S2S models. To eliminate this restriction on research possibilities such as modeling unit choice, we present RASR2 in this work, a research-oriented generic S2S decoder implemented in C++. It offers a strong flexibility/compatibility for various S2S models, language models, label units/topologies and neural network architectures. It provides efficient decoding for both open- and closed-vocabulary scenarios based on a generalized search framework with rich support for different search modes and settings. We evaluate RASR2 with a wide range of experiments on both switchboard and Librispeech corpora. Our source code is public online.
Recently, graph-based and Transformer-based deep learning networks have demonstrated excellent performances on various point cloud tasks. Most of the existing graph methods are based on static graph, which take a fixed input to establish graph relations. Moreover, many graph methods apply maximization and averaging to aggregate neighboring features, so that only a single neighboring point affects the feature of centroid or different neighboring points have the same influence on the centroid's feature, which ignoring the correlation and difference between points. Most Transformer-based methods extract point cloud features based on global attention and lack the feature learning on local neighbors. To solve the problems of these two types of models, we propose a new feature extraction block named Graph Transformer and construct a 3D point point cloud learning network called GTNet to learn features of point clouds on local and global patterns. Graph Transformer integrates the advantages of graph-based and Transformer-based methods, and consists of Local Transformer and Global Transformer modules. Local Transformer uses a dynamic graph to calculate all neighboring point weights by intra-domain cross-attention with dynamically updated graph relations, so that every neighboring point could affect the features of centroid with different weights; Global Transformer enlarges the receptive field of Local Transformer by a global self-attention. In addition, to avoid the disappearance of the gradient caused by the increasing depth of network, we conduct residual connection for centroid features in GTNet; we also adopt the features of centroid and neighbors to generate the local geometric descriptors in Local Transformer to strengthen the local information learning capability of the model. Finally, we use GTNet for shape classification, part segmentation and semantic segmentation tasks in this paper.
Recently, Transformer-based methods for point cloud learning have achieved good results on various point cloud learning benchmarks. However, since the attention mechanism needs to generate three feature vectors of query, key, and value to calculate attention features, most of the existing Transformer-based point cloud learning methods usually consume a large amount of computational time and memory resources when calculating global attention. To address this problem, we propose a Voxel-Transformer-Point (VTP) Block for extracting local and global features of point clouds. VTP combines the advantages of voxel-based, point-based and Transformer-based methods, which consists of Voxel-Based Branch (V branch), Point-Based Transformer Branch (PT branch) and Point-Based Branch (P branch). The V branch extracts the coarse-grained features of the point cloud through low voxel resolution; the PT branch obtains the fine-grained features of the point cloud by calculating the self-attention in the local neighborhood and the inter-neighborhood cross-attention; the P branch uses a simplified MLP network to generate the global location information of the point cloud. In addition, to enrich the local features of point clouds at different scales, we set the voxel scale in the V branch and the neighborhood sphere scale in the PT branch to one large and one small (large voxel scale \& small neighborhood sphere scale or small voxel scale \& large neighborhood sphere scale). Finally, we use VTP as the feature extraction network to construct a VTPNet for point cloud learning, and performs shape classification, part segmentation, and semantic segmentation tasks on the ModelNet40, ShapeNet Part, and S3DIS datasets. The experimental results indicate that VTPNet has good performance in 3D point cloud learning.
Existing data-to-text generation efforts mainly focus on generating a coherent text from non-linguistic input data, such as tables and attribute-value pairs, but overlook that different application scenarios may require texts of different styles. Inspired by this, we define a new task, namely stylized data-to-text generation, whose aim is to generate coherent text for the given non-linguistic data according to a specific style. This task is non-trivial, due to three challenges: the logic of the generated text, unstructured style reference, and biased training samples. To address these challenges, we propose a novel stylized data-to-text generation model, named StyleD2T, comprising three components: logic planning-enhanced data embedding, mask-based style embedding, and unbiased stylized text generation. In the first component, we introduce a graph-guided logic planner for attribute organization to ensure the logic of generated text. In the second component, we devise feature-level mask-based style embedding to extract the essential style signal from the given unstructured style reference. In the last one, pseudo triplet augmentation is utilized to achieve unbiased text generation, and a multi-condition based confidence assignment function is designed to ensure the quality of pseudo samples. Extensive experiments on a newly collected dataset from Taobao have been conducted, and the results show the superiority of our model over existing methods.
Existing neural methods have shown great potentials towards generating informative text from structured tabular data as well as maintaining high content fidelity. However, few of them shed light on generating personalized expressions, which often requires well-aligned persona-table-text datasets that are difficult to obtain. To overcome these obstacles, we explore personalized table-to-text generation under a zero-shot setting, by assuming no well-aligned persona-table-text triples are required during training. To this end, we firstly collect a set of unpaired persona information and then propose a semi-supervised approach with contrastive persona distillation (S2P-CPD) to generate personalized context. Specifically, tabular data and persona information are firstly represented as latent variables separately. Then, we devise a latent space fusion technique to distill persona information into the table representation. Besides, a contrastive-based discriminator is employed to guarantee the style consistency between the generated context and its corresponding persona. Experimental results on two benchmarks demonstrate S2P-CPD's ability on keeping both content fidelity and personalized expressions.
Weakly-supervised temporal action localization aims to localize action instances in untrimmed videos with only video-level supervision. We witness that different actions record common phases, e.g., the run-up in the HighJump and LongJump. These different actions are defined as conjoint actions, whose rest parts are definite phases, e.g., leaping over the bar in a HighJump. Compared with the common phases, the definite phases are more easily localized in existing researches. Most of them formulate this task as a Multiple Instance Learning paradigm, in which the common phases are tended to be confused with the background, and affect the localization completeness of the conjoint actions. To tackle this challenge, we propose a Joint of Common and Definite phases Network (JCDNet) by improving feature discriminability of the conjoint actions. Specifically, we design a Class-Aware Discriminative module to enhance the contribution of the common phases in classification by the guidance of the coarse definite-phase features. Besides, we introduce a temporal attention module to learn robust action-ness scores via modeling temporal dependencies, distinguishing the common phases from the background. Extensive experiments on three datasets (THUMOS14, ActivityNetv1.2, and a conjoint-action subset) demonstrate that JCDNet achieves competitive performance against the state-of-the-art methods. Keywords: weakly-supervised learning, temporal action localization, conjoint action
Reconstructing perceived natural images or decoding their categories from fMRI signals are challenging tasks with great scientific significance. Due to the lack of paired samples, most existing methods fail to generate semantically recognizable reconstruction and are difficult to generalize to novel classes. In this work, we propose, for the first time, a task-agnostic brain decoding model by unifying the visual stimulus classification and reconstruction tasks in a semantic space. We denote it as BrainCLIP, which leverages CLIP's cross-modal generalization ability to bridge the modality gap between brain activities, images, and texts. Specifically, BrainCLIP is a VAE-based architecture that transforms fMRI patterns into the CLIP embedding space by combining visual and textual supervision. Note that previous works rarely use multi-modal supervision for visual stimulus decoding. Our experiments demonstrate that textual supervision can significantly boost the performance of decoding models compared to the condition where only image supervision exists. BrainCLIP can be applied to multiple scenarios like fMRI-to-image generation, fMRI-image-matching, and fMRI-text-matching. Compared with BraVL, a recently proposed multi-modal method for fMRI-based brain decoding, BrainCLIP achieves significantly better performance on the novel class classification task. BrainCLIP also establishes a new state-of-the-art for fMRI-based natural image reconstruction in terms of high-level image features.
Omnidirectional image quality assessment (OIQA) aims to predict the perceptual quality of omnidirectional images that cover the whole 180$\times$360$^{\circ}$ viewing range of the visual environment. Here we propose a blind/no-reference OIQA method named S$^2$ that bridges the gap between low-level statistics and high-level semantics of omnidirectional images. Specifically, statistic and semantic features are extracted in separate paths from multiple local viewports and the hallucinated global omnidirectional image, respectively. A quality regression along with a weighting process is then followed that maps the extracted quality-aware features to a perceptual quality prediction. Experimental results demonstrate that the proposed S$^2$ method offers highly competitive performance against state-of-the-art methods.
Reconstructing a 3D object from a 2D image is a well-researched vision problem, with many kinds of deep learning techniques having been tried. Most commonly, 3D convolutional approaches are used, though previous work has shown state-of-the-art methods using 2D convolutions that are also significantly more efficient to train. With the recent rise of transformers for vision tasks, often outperforming convolutional methods, along with some earlier attempts to use transformers for 3D object reconstruction, we set out to use visual transformers in place of convolutions in existing efficient, high-performing techniques for 3D object reconstruction in order to achieve superior results on the task. Using a transformer-based encoder and decoder to predict 3D structure from 2D images, we achieve accuracy similar or superior to the baseline approach. This study serves as evidence for the potential of visual transformers in the task of 3D object reconstruction.
Reward design for reinforcement learning agents can be difficult in situations where one not only wants the agent to achieve some effect in the world but where one also cares about how that effect is achieved. For example, we might wish for an agent to adhere to a tacit understanding of commonsense, align itself to a preference for how to behave for purposes of safety, or taking on a particular role in an interactive game. Storytelling is a mode for communicating tacit procedural knowledge. We introduce a technique, Story Shaping, in which a reinforcement learning agent infers tacit knowledge from an exemplar story of how to accomplish a task and intrinsically rewards itself for performing actions that make its current environment adhere to that of the inferred story world. Specifically, Story Shaping infers a knowledge graph representation of the world state from observations, and also infers a knowledge graph from the exemplar story. An intrinsic reward is generated based on the similarity between the agent's inferred world state graph and the inferred story world graph. We conducted experiments in text-based games requiring commonsense reasoning and shaping the behaviors of agents as virtual game characters.