Weakly supervised named entity recognition methods train label models to aggregate the token annotations of multiple noisy labeling functions (LFs) without seeing any manually annotated labels. To work well, the label model needs to contextually identify and emphasize well-performed LFs while down-weighting the under-performers. However, evaluating the LFs is challenging due to the lack of ground truths. To address this issue, we propose the sparse conditional hidden Markov model (Sparse-CHMM). Instead of predicting the entire emission matrix as other HMM-based methods, Sparse-CHMM focuses on estimating its diagonal elements, which are considered as the reliability scores of the LFs. The sparse scores are then expanded to the full-fledged emission matrix with pre-defined expansion functions. We also augment the emission with weighted XOR scores, which track the probabilities of an LF observing incorrect entities. Sparse-CHMM is optimized through unsupervised learning with a three-stage training pipeline that reduces the training difficulty and prevents the model from falling into local optima. Compared with the baselines in the Wrench benchmark, Sparse-CHMM achieves a 3.01 average F1 score improvement on five comprehensive datasets. Experiments show that each component of Sparse-CHMM is effective, and the estimated LF reliabilities strongly correlate with true LF F1 scores.
Contextual knowledge is essential for reducing speech recognition errors on high-valued long-tail words. This paper proposes a novel tree-constrained pointer generator (TCPGen) component that enables end-to-end ASR models to bias towards a list of long-tail words obtained using external contextual information. With only a small overhead in memory use and computation cost, TCPGen can structure thousands of biasing words efficiently into a symbolic prefix-tree and creates a neural shortcut between the tree and the final ASR output to facilitate the recognition of the biasing words. To enhance TCPGen, we further propose a novel minimum biasing word error (MBWE) loss that directly optimises biasing word errors during training, along with a biasing-word-driven language model discounting (BLMD) method during the test. All contextual ASR systems were evaluated on the public Librispeech audiobook corpus and the data from the dialogue state tracking challenges (DSTC) with the biasing lists extracted from the dialogue-system ontology. Consistent word error rate (WER) reductions were achieved with TCPGen, which were particularly significant on the biasing words with around 40\% relative reductions in the recognition error rates. MBWE and BLMD further improved the effectiveness of TCPGen and achieved more significant WER reductions on the biasing words. TCPGen also achieved zero-shot learning of words not in the audio training set with large WER reductions on the out-of-vocabulary words in the biasing list.
Physics-informed neural networks (PINNs) provide a deep learning framework for numerically solving partial differential equations (PDEs), and have been widely used in a variety of PDE problems. However, there still remain some challenges in the application of PINNs: 1) the mechanism of PINNs is unsuitable (at least cannot be directly applied) to exploiting a small size of (usually very few) extra informative samples to refine the networks; and 2) the efficiency of training PINNs often becomes low for some complicated PDEs. In this paper, we propose the generative adversarial physics-informed neural network (GA-PINN), which integrates the generative adversarial (GA) mechanism with the structure of PINNs, to improve the performance of PINNs by exploiting only a small size of exact solutions to the PDEs. Inspired from the weighting strategy of the Adaboost method, we then introduce a point-weighting (PW) method to improve the training efficiency of PINNs, where the weight of each sample point is adaptively updated at each training iteration. The numerical experiments show that GA-PINNs outperform PINNs in many well-known PDEs and the PW method also improves the efficiency of training PINNs and GA-PINNs.
The new concept of semi-integrated-sensing-and-communication (Semi-ISaC) is proposed for next-generation cellular networks. We propose a novel Semi-ISaC framework which provides more freedom as it allows that a portion of the bandwidth is exclusively used for either wireless communication or for radar detection, while the rest is for ISaC transmission. To enhance the bandwidth efficiency (BE), we investigate the evolution of Semi-ISaC networks from orthogonal multiple access (OMA) to non-orthogonal multiple access (NOMA). First, we evaluate the performance of an OMA-based Semi-ISaC network. As for the communication signals, we investigate both the outage probability (OP) and the ergodic rate. As for the radar echoes, we characterize the ergodic radar estimation information rate (REIR). Based on these metrics, we derive the analytical and asymptotic expressions of the ergodic RIER. Then, we investigate the performance of a NOMA-based Semi-ISaC network. More specifically, we derive the analytical expressions of both the OP and of the ergodic rate for the communication signals. The asymptotic expressions of OP are also derived for quantifying the diversity gains of the communication signals. As for the radar echoes, we derive the analytical and asymptotic ergodic REIR. The high signal-to-noise ratio (SNR) slopes are also evaluated. The analytical results indicate that: 1) Under a two-user NOMA Semi-ISaC scenario, the diversity order of the near-user equals to the coefficient of the Nakagami-m fading channels ($m$), while that of the far-user is zero; and 2) The high-SNR slope for the ergodic REIR is based on the ratio of the radar signal's duty cycle and the pulse duration. Our simulation results show that: 1) Semi-ISaC has better channel capacity than conventional ISaC; and 2) NOMA-based Semi-ISaC has better channel capacity than OMA-based Semi-ISaC.
User sessions empower many search and recommendation tasks on a daily basis. Such session data are semi-structured, which encode heterogeneous relations between queries and products, and each item is described by the unstructured text. Despite recent advances in self-supervised learning for text or graphs, there lack of self-supervised learning models that can effectively capture both intra-item semantics and inter-item interactions for semi-structured sessions. To fill this gap, we propose CERES, a graph-based transformer model for semi-structured session data. CERES learns representations that capture both inter- and intra-item semantics with (1) a graph-conditioned masked language pretraining task that jointly learns from item text and item-item relations; and (2) a graph-conditioned transformer architecture that propagates inter-item contexts to item-level representations. We pretrained CERES using ~468 million Amazon sessions and find that CERES outperforms strong pretraining baselines by up to 9% in three session search and entity linking tasks.
We introduce Optical Flow TransFormer (FlowFormer), a transformer-based neural network architecture for learning optical flow. FlowFormer tokenizes the 4D cost volume built from an image pair, encodes the cost tokens into a cost memory with alternate-group transformer (AGT) layers in a novel latent space, and decodes the cost memory via a recurrent transformer decoder with dynamic positional cost queries. On the Sintel benchmark clean pass, FlowFormer achieves 1.178 average end-ponit-error (AEPE), a 15.1% error reduction from the best published result (1.388). Besides, FlowFormer also achieves strong generalization performance. Without being trained on Sintel, FlowFormer achieves 1.00 AEPE on the Sintel training set clean pass, outperforming the best published result (1.29) by 22.4%.
Unsupervised point cloud completion aims at estimating the corresponding complete point cloud of a partial point cloud in an unpaired manner. It is a crucial but challenging problem since there is no paired partial-complete supervision that can be exploited directly. In this work, we propose a novel framework, which learns a unified and structured latent space that encoding both partial and complete point clouds. Specifically, we map a series of related partial point clouds into multiple complete shape and occlusion code pairs and fuse the codes to obtain their representations in the unified latent space. To enforce the learning of such a structured latent space, the proposed method adopts a series of constraints including structured ranking regularization, latent code swapping constraint, and distribution supervision on the related partial point clouds. By establishing such a unified and structured latent space, better partial-complete geometry consistency and shape completion accuracy can be achieved. Extensive experiments show that our proposed method consistently outperforms state-of-the-art unsupervised methods on both synthetic ShapeNet and real-world KITTI, ScanNet, and Matterport3D datasets.
Low-light image enhancement - a pervasive but challenging problem, plays a central role in enhancing the visibility of an image captured in a poor illumination environment. Due to the fact that not all photons can pass the Bayer-Filter on the sensor of the color camera, in this work, we first present a De-Bayer-Filter simulator based on deep neural networks to generate a monochrome raw image from the colored raw image. Next, a fully convolutional network is proposed to achieve the low-light image enhancement by fusing colored raw data with synthesized monochrome raw data. Channel-wise attention is also introduced to the fusion process to establish a complementary interaction between features from colored and monochrome raw images. To train the convolutional networks, we propose a dataset with monochrome and color raw pairs named Mono-Colored Raw paired dataset (MCR) collected by using a monochrome camera without Bayer-Filter and a color camera with Bayer-Filter. The proposed pipeline take advantages of the fusion of the virtual monochrome and the color raw images and our extensive experiments indicate that significant improvement can be achieved by leveraging raw sensor data and data-driven learning.