In cross-lingual language models, representations for many different languages live in the same space. Here, we investigate the linguistic and non-linguistic factors affecting sentence-level alignment in cross-lingual pretrained language models for 1
01 languages and 5,050 language pairs. Using BERT-based LaBSE and BiLSTM-based LASER as our models, and the Bible as our corpus, we compute a task-based measure of cross-lingual alignment in the form of bitext retrieval performance, as well as four intrinsic measures of vector space alignment and isomorphism. We then examine a range of linguistic, quasi-linguistic, and training-related features as potential predictors of these alignment metrics. The results of our analyses show that word order agreement and agreement in morphological complexity are two of the strongest linguistic predictors of cross-linguality. We also note in-family training data as a stronger predictor than language-specific training data across the board. We verify some of our linguistic findings by looking at the effect of morphological segmentation on English-Inuktitut alignment, in addition to examining the effect of word order agreement on isomorphism for 66 zero-shot language pairs from a different corpus. We make the data and code for our experiments publicly available.
Multi-task learning (MTL) is an efficient way to improve the performance of related tasks by sharing knowledge. However, most existing MTL networks run on a single end and are not suitable for collaborative intelligence (CI) scenarios. In this work,
we propose an MTL network with a deep joint source-channel coding (JSCC) framework, which allows operating under CI scenarios. We first propose a feature fusion based MTL network (FFMNet) for joint object detection and semantic segmentation. Compared with other MTL networks, FFMNet gets higher performance with fewer parameters. Then FFMNet is split into two parts, which run on a mobile device and an edge server respectively. The feature generated by the mobile device is transmitted through the wireless channel to the edge server. To reduce the transmission overhead of the intermediate feature, a deep JSCC network is designed. By combining two networks together, the whole model achieves 512 times compression for the intermediate feature and a performance loss within 2% on both tasks. At last, by training with noise, the FFMNet with JSCC is robust to various channel conditions and outperforms the separate source and channel coding scheme.
Video question answering (VideoQA) is challenging given its multimodal combination of visual understanding and natural language understanding. While existing approaches seldom leverage the appearance-motion information in the video at multiple tempor
al scales, the interaction between the question and the visual information for textual semantics extraction is frequently ignored. Targeting these issues, this paper proposes a novel Temporal Pyramid Transformer (TPT) model with multimodal interaction for VideoQA. The TPT model comprises two modules, namely Question-specific Transformer (QT) and Visual Inference (VI). Given the temporal pyramid constructed from a video, QT builds the question semantics from the coarse-to-fine multimodal co-occurrence between each word and the visual content. Under the guidance of such question-specific semantics, VI infers the visual clues from the local-to-global multi-level interactions between the question and the video. Within each module, we introduce a multimodal attention mechanism to aid the extraction of question-video interactions, with residual connections adopted for the information passing across different levels. Through extensive experiments on three VideoQA datasets, we demonstrate better performances of the proposed method in comparison with the state-of-the-arts.
The annotation of domain experts is important for some medical applications where the objective groundtruth is ambiguous to define, e.g., the rehabilitation for some chronic diseases, and the prescreening of some musculoskeletal abnormalities without
further medical examinations. However, improper uses of the annotations may hinder developing reliable models. On one hand, forcing the use of a single groundtruth generated from multiple annotations is less informative for the modeling. On the other hand, feeding the model with all the annotations without proper regularization is noisy given existing disagreements. For such issues, we propose a novel agreement learning framework to tackle the challenge of learning from multiple annotators without objective groundtruth. The framework has two streams, with one stream fitting with the multiple annotators and the other stream learning agreement information between the annotators. In particular, the agreement learning stream produces regularization information to the classifier stream, tuning its decision to be better in line with the agreement between the annotators. The proposed method can be easily plugged to existing backbones developed with majority-voted groundtruth or multiple annotations. Thereon, experiments on two medical datasets demonstrate improved agreement levels with annotators.
High-resolution (HR) hyperspectral face image plays an important role in face related computer vision tasks under uncontrolled conditions, such as low-light environment and spoofing attacks. However, the dense spectral bands of hyperspectral face ima
ges come at the cost of limited amount of photons reached a narrow spectral window on average, which greatly reduces the spatial resolution of hyperspectral face images. In this paper, we investigate how to adapt the deep learning techniques to hyperspectral face image super-resolution (HFSR), especially when the training samples are very limited. Benefiting from the amount of spectral bands, in which each band can be seen as an image, we present a spectral splitting and aggregation network (SSANet) for HFSR with limited training samples. In the shallow layers, we split the hyperspectral image into different spectral groups and take each of them as an individual training sample (in the sense that each group will be fed into the same network). Then, we gradually aggregate the neighbor bands at the deeper layers to exploit the spectral correlations. By this spectral splitting and aggregation strategy (SSAS), we can divide the original hyperspectral image into multiple samples to support the efficient training of the network and effectively exploit the spectral correlations among spectrum. To cope with the challenge of small training sample size (S3) problem, we propose to expand the training samples by a self-representation model and symmetry-induced augmentation. Experiments show that the introduced SSANet can well model the joint correlations of spatial and spectral information. By expanding the training samples, our proposed method can effectively alleviate the S3 problem. The comparison results demonstrate that our proposed method can outperform the state-of-the-arts.
Text-to-speech and co-speech gesture synthesis have until now been treated as separate areas by two different research communities, and applications merely stack the two technologies using a simple system-level pipeline. This can lead to modeling ine
fficiencies and may introduce inconsistencies that limit the achievable naturalness. We propose to instead synthesize the two modalities in a single model, a new problem we call integrated speech and gesture synthesis (ISG). We also propose a set of models modified from state-of-the-art neural speech-synthesis engines to achieve this goal. We evaluate the models in three carefully-designed user studies, two of which evaluate the synthesized speech and gesture in isolation, plus a combined study that evaluates the models like they will be used in real-world applications -- speech and gesture presented together. The results show that participants rate one of the proposed integrated synthesis models as being as good as the state-of-the-art pipeline system we compare against, in all three tests. The model is able to achieve this with faster synthesis time and greatly reduced parameter count compared to the pipeline system, illustrating some of the potential benefits of treating speech and gesture synthesis together as a single, unified problem. Videos and code are available on our project page at https://swatsw.github.io/isg_icmi21/
The anchor-based detectors handle the problem of scale variation by building the feature pyramid and directly setting different scales of anchors on each cell in different layers. However, it is difficult for box-wise anchors to guide the adaptive le
arning of scale-specific features in each layer because there is no one-to-one correspondence between box-wise anchors and pixel-level features. In order to alleviate the problem, in this paper, we propose a scale-customized weak segmentation (SCWS) block at the pixel level for scale customized object feature learning in each layer. By integrating the SCWS blocks into the single-shot detector, a scale-aware object detector (SCOD) is constructed to detect objects of different sizes naturally and accurately. Furthermore, the standard location loss neglects the fact that the hard and easy samples may be seriously imbalanced. A forthcoming problem is that it is unable to get more accurate bounding boxes due to the imbalance. To address this problem, an adaptive IoU (AIoU) loss via a simple yet effective squeeze operation is specified in our SCOD. Extensive experiments on PASCAL VOC and MS COCO demonstrate the superiority of our SCOD.
Classification and regression are two pillars of object detectors. In most CNN-based detectors, these two pillars are optimized independently. Without direct interactions between them, the classification loss and the regression loss can not be optimi
zed synchronously toward the optimal direction in the training phase. This clearly leads to lots of inconsistent predictions with high classification score but low localization accuracy or low classification score but high localization accuracy in the inference phase, especially for the objects of irregular shape and occlusion, which severely hurts the detection performance of existing detectors after NMS. To reconcile prediction consistency for balanced object detection, we propose a Harmonic loss to harmonize the optimization of classification branch and localization branch. The Harmonic loss enables these two branches to supervise and promote each other during training, thereby producing consistent predictions with high co-occurrence of top classification and localization in the inference phase. Furthermore, in order to prevent the localization loss from being dominated by outliers during training phase, a Harmonic IoU loss is proposed to harmonize the weight of the localization loss of different IoU-level samples. Comprehensive experiments on benchmarks PASCAL VOC and MS COCO demonstrate the generality and effectiveness of our model for facilitating existing object detectors to state-of-the-art accuracy.
It is well known that on the surface of Weyl semimetals, Fermi arcs appear as the topologically protected surface states. In this work, we give a semiclassical explanation for the morphology of the surface Fermi arcs. Viewing the surface states as a
two-dimensional Fermi gas subject to band bending and Berry curvatures, we show that it is the non-parallelism between the velocity and the momentum that gives rise to the spiraling Fermi arcs. We map out the Fermi arcs from the velocity field for a single Weyl point and a lattice with two Weyl points. We also investigate the surface magnetoplasma of Dirac semimetals in a magnetic field. In this case, the surface states obtains chiral nature from both drift motion and the chiral magnetic effect, resulting in Fermi arcs. We also discuss the important role played by the Imbert-Fedorov shift in the formation of surface Fermi arcs.
We designed and demonstrated experimentally a silicon photonics integrated dynamic polarization controller which is a crucial component of a continuous-variable quantum key distribution system. By using a variable step simulated annealing approach, w
e achieve a dynamic polarization extinction ratio greater than 25 dB. The dynamic polarization controller can be utilized in silicon photonics integrated continuous-variable quantum key distribution system to minimize the size and decrease the cost further.
