With the recent success of pre-trained models in NLP, a significant focus was put on interpreting their representations. One of the most prominent approaches is structural probing (Hewitt and Manning, 2019), where a linear projection of word embeddings is performed in order to approximate the topology of dependency structures. In this work, we introduce a new type of structural probing, where the linear projection is decomposed into 1. isomorphic space rotation; 2. linear scaling that identifies and scales the most relevant dimensions. In addition to syntactic dependency, we evaluate our method on novel tasks (lexical hypernymy and position in a sentence). We jointly train the probes for multiple tasks and experimentally show that lexical and syntactic information is separated in the representations. Moreover, the orthogonal constraint makes the Structural Probes less vulnerable to memorization.
Sindhi word segmentation is a challenging task due to space omission and insertion issues. The Sindhi language itself adds to this complexity. It's cursive and consists of characters with inherent joining and non-joining properties, independent of word boundaries. Existing Sindhi word segmentation methods rely on designing and combining hand-crafted features. However, these methods have limitations, such as difficulty handling out-of-vocabulary words, limited robustness for other languages, and inefficiency with large amounts of noisy or raw text. Neural network-based models, in contrast, can automatically capture word boundary information without requiring prior knowledge. In this paper, we propose a Subword-Guided Neural Word Segmenter (SGNWS) that addresses word segmentation as a sequence labeling task. The SGNWS model incorporates subword representation learning through a bidirectional long short-term memory encoder, position-aware self-attention, and a conditional random field. Our empirical results demonstrate that the SGNWS model achieves state-of-the-art performance in Sindhi word segmentation on six datasets.
Self attention networks (SANs) have been widely utilized in recent NLP studies. Unlike CNNs or RNNs, standard SANs are usually position-independent, and thus are incapable of capturing the structural priors between sequences of words. Existing studies commonly apply one single mask strategy on SANs for incorporating structural priors while failing at modeling more abundant structural information of texts. In this paper, we aim at introducing multiple types of structural priors into SAN models, proposing the Multiple Structural Priors Guided Self Attention Network (MS-SAN) that transforms different structural priors into different attention heads by using a novel multi-mask based multi-head attention mechanism. In particular, we integrate two categories of structural priors, including the sequential order and the relative position of words. For the purpose of capturing the latent hierarchical structure of the texts, we extract these information not only from the word contexts but also from the dependency syntax trees. Experimental results on two tasks show that MS-SAN achieves significant improvements against other strong baselines.
In genetic networks, information of relevance to the organism is represented by the concentrations of transcription factor molecules. In order to extract this information the cell must effectively "measure"' these concentrations, but there are physical limits to the precision of these measurements. We explore this trading between bits of precision in measuring concentration and bits of relevant information that can be extracted, using the gap gene network in the early fly embryo as an example. We argue that cells in the embryo can extract all the available information about their position, but only if the concentration measurements approach the physical limits to information capacity. These limits necessitate the observed proliferation of enhancer elements with sensitivities to combinations of transcription factors, but fine tuning of the parameters of these multiple enhancers is not required.
Increasing the input length has been a driver of progress in language modeling with transformers. We identify conditions where shorter inputs are not harmful, and achieve perplexity and efficiency improvements through two new methods that decrease input length. First, we show that initially training a model on short subsequences before moving on to longer ones both reduces overall training time and, surprisingly, substantially improves perplexity. Second, we show how to improve the efficiency of recurrence methods in transformers, which let models condition on previously processed tokens when generating sequences that exceed the maximal length the transformer can handle at once. Existing methods require computationally expensive relative position embeddings; we introduce a simple alternative of adding absolute position embeddings to queries and keys instead of to word embeddings, which efficiently produces superior results. We show that these recurrent models also benefit from short input lengths. Combining these techniques speeds up training by a factor of 1.65, reduces memory usage, and substantially improves perplexity on WikiText-103, without adding any parameters.
Adversarial examples have gained tons of attention in recent years. Many adversarial attacks have been proposed to attack image classifiers, but few work shift attention to object detectors. In this paper, we propose Sparse Adversarial Attack (SAA) which enables adversaries to perform effective evasion attack on detectors with bounded \emph{l$_{0}$} norm perturbation. We select the fragile position of the image and designed evasion loss function for the task. Experiment results on YOLOv4 and FasterRCNN reveal the effectiveness of our method. In addition, our SAA shows great transferability across different detectors in the black-box attack setting. Codes are available at \emph{https://github.com/THUrssq/Tianchi04}.
In this paper, we present Wi-Mose, the first 3D moving human pose estimation system using commodity WiFi. Previous WiFi-based works have achieved 2D and 3D pose estimation. These solutions either capture poses from one perspective or construct poses of people who are at a fixed point, preventing their wide adoption in daily scenarios. To reconstruct 3D poses of people who move throughout the space rather than a fixed point, we fuse the amplitude and phase into Channel State Information (CSI) images which can provide both pose and position information. Besides, we design a neural network to extract features that are only associated with poses from CSI images and then convert the features into key-point coordinates. Experimental results show that Wi-Mose can localize key-point with 29.7mm and 37.8mm Procrustes analysis Mean Per Joint Position Error (P-MPJPE) in the Line of Sight (LoS) and Non-Line of Sight (NLoS) scenarios, respectively, achieving higher performance than the state-of-the-art method. The results indicate that Wi-Mose can capture high-precision 3D human poses throughout the space.
In this report we provide an improvement of the significance adjustment from the FA*IR algorithm of Zehlike et al., which did not work for very short rankings in combination with a low minimum proportion $p$ for the protected group. We show how the minimum number of protected candidates per ranking position can be calculated exactly and provide a mapping from the continuous space of significance levels ($\alpha$) to a discrete space of tables, which allows us to find $\alpha_c$ using a binary search heuristic.
A smart vehicle should be able to understand human behavior and predict their actions to avoid hazardous situations. Specific traits in human behavior can be automatically predicted, which can help the vehicle make decisions, increasing safety. One of the most important aspects pertaining to the driving task is the driver's visual attention. Predicting the driver's visual attention can help a vehicle understand the awareness state of the driver, providing important contextual information. While estimating the exact gaze direction is difficult in the car environment, a coarse estimation of the visual attention can be obtained by tracking the position and orientation of the head. Since the relation between head pose and gaze direction is not one-to-one, this paper proposes a formulation based on probabilistic models to create salient regions describing the visual attention of the driver. The area of the predicted region is small when the model has high confidence on the prediction, which is directly learned from the data. We use Gaussian process regression (GPR) to implement the framework, comparing the performance with different regression formulations such as linear regression and neural network based methods. We evaluate these frameworks by studying the tradeoff between spatial resolution and accuracy of the probability map using naturalistic recordings collected with the UTDrive platform. We observe that the GPR method produces the best result creating accurate predictions with localized salient regions. For example, the 95% confidence region is defined by an area that covers 3.77% region of a sphere surrounding the driver.
A common approach to modeling networks assigns each node to a position on a low-dimensional manifold where distance is inversely proportional to connection likelihood. More positive manifold curvature encourages more and tighter communities; negative curvature induces repulsion. We consistently estimate manifold type, dimension, and curvature from simply connected, complete Riemannian manifolds of constant curvature. We represent the graph as a noisy distance matrix based on the ties between cliques, then develop hypothesis tests to determine whether the observed distances could plausibly be embedded isometrically in each of the candidate geometries. We apply our approach to data-sets from economics and neuroscience.
Fall detection (FD) systems are important assistive technologies for healthcare that can detect emergency fall events and alert caregivers. However, it is not easy to obtain large-scale annotated fall events with various specifications of sensors or sensor positions during the implementation of accurate FD systems. Moreover, the knowledge obtained through machine learning has been restricted to tasks in the same domain. The mismatch between different domains might hinder the performance of FD systems. Cross-domain knowledge transfer is very beneficial for machine-learning-based FD systems to train a reliable FD model with well-labeled data in new environments. In this study, we propose domain-adaptive fall detection (DAFD) using deep adversarial training (DAT) to tackle cross-domain problems, such as cross-position and cross-configuration. The proposed DAFD can transfer knowledge from the source domain to the target domain by minimizing the domain discrepancy to avoid mismatch problems. The experimental results show that the average F1-score improvement when using DAFD ranges from 1.5% to 7% in the cross-position scenario, and from 3.5% to 12% in the cross-configuration scenario, compared to using the conventional FD model without domain adaptation training. The results demonstrate that the proposed DAFD successfully helps to deal with cross-domain problems and to achieve better detection performance.
Recent work on audio-visual navigation assumes a constantly-sounding target and restricts the role of audio to signaling the target's position. We introduce semantic audio-visual navigation, where objects in the environment make sounds consistent with their semantic meaning (e.g., toilet flushing, door creaking) and acoustic events are sporadic or short in duration. We propose a transformer-based model to tackle this new semantic AudioGoal task, incorporating an inferred goal descriptor that captures both spatial and semantic properties of the target. Our model's persistent multimodal memory enables it to reach the goal even long after the acoustic event stops. In support of the new task, we also expand the SoundSpaces audio simulations to provide semantically grounded sounds for an array of objects in Matterport3D. Our method strongly outperforms existing audio-visual navigation methods by learning to associate semantic, acoustic, and visual cues.
Non-local operations are usually used to capture long-range dependencies via aggregating global context to each position recently. However, most of the methods cannot preserve object shapes since they only focus on feature similarity but ignore proximity between central and other positions for capturing long-range dependencies, while shape-awareness is beneficial to many computer vision tasks. In this paper, we propose a Semi-Global Shape-aware Network (SGSNet) considering both feature similarity and proximity for preserving object shapes when modeling long-range dependencies. A hierarchical way is taken to aggregate global context. In the first level, each position in the whole feature map only aggregates contextual information in vertical and horizontal directions according to both similarity and proximity. And then the result is input into the second level to do the same operations. By this hierarchical way, each central position gains supports from all other positions, and the combination of similarity and proximity makes each position gain supports mostly from the same semantic object. Moreover, we also propose a linear time algorithm for the aggregation of contextual information, where each of rows and columns in the feature map is treated as a binary tree to reduce similarity computation cost. Experiments on semantic segmentation and image retrieval show that adding SGSNet to existing networks gains solid improvements on both accuracy and efficiency.
Accurately predicting the binding affinity between drugs and proteins is an essential step for computational drug discovery. Since graph neural networks (GNNs) have demonstrated remarkable success in various graph-related tasks, GNNs have been considered as a promising tool to improve the binding affinity prediction in recent years. However, most of the existing GNN architectures can only encode the topological graph structure of drugs and proteins without considering the relative spatial information among their atoms. Whereas, different from other graph datasets such as social networks and commonsense knowledge graphs, the relative spatial position and chemical bonds among atoms have significant impacts on the binding affinity. To this end, in this paper, we propose a diStance-aware Molecule graph Attention Network (S-MAN) tailored to drug-target binding affinity prediction. As a dedicated solution, we first propose a position encoding mechanism to integrate the topological structure and spatial position information into the constructed pocket-ligand graph. Moreover, we propose a novel edge-node hierarchical attentive aggregation structure which has edge-level aggregation and node-level aggregation. The hierarchical attentive aggregation can capture spatial dependencies among atoms, as well as fuse the position-enhanced information with the capability of discriminating multiple spatial relations among atoms. Finally, we conduct extensive experiments on two standard datasets to demonstrate the effectiveness of S-MAN.
The placement of text over an image is an important part of producing high-quality visual designs. Automating this work by determining appropriate position, orientation, and style for textual elements requires understanding the contents of the background image. We refer to the search for aesthetic parameters of text rendered over images as "copyspace detection", noting that this task is distinct from foreground-background separation. We have developed solutions using one and two stage object detection methodologies trained on an expertly labeled data. This workshop will examine such algorithms for copyspace detection and demonstrate their application in generative design models and pipelines such as Einstein Designer.
Recent papers have shown that sufficiently overparameterized neural networks can perfectly fit even random labels. Thus, it is crucial to understand the underlying reason behind the generalization performance of a network on real-world data. In this work, we propose several measures to predict the generalization error of a network given the training data and its parameters. Using one of these measures, based on noise resilience of the network, we secured 5th position in the predicting generalization in deep learning (PGDL) competition at NeurIPS 2020.
Deep learning networks have shown promising performance for accurate object localization in medial images, but require large amount of annotated data for supervised training, which is expensive and expertise burdensome. To address this problem, we present a one-shot framework for organ and landmark localization in volumetric medical images, which does not need any annotation during the training stage and could be employed to locate any landmarks or organs in test images given a support (reference) image during the inference stage. Our main idea comes from that tissues and organs from different human bodies have a similar relative position and context. Therefore, we could predict the relative positions of their non-local patches, thus locate the target organ. Our framework is composed of three parts: (1) A projection network trained to predict the 3D offset between any two patches from the same volume, where human annotations are not required. In the inference stage, it takes one given landmark in a reference image as a support patch and predicts the offset from a random patch to the corresponding landmark in the test (query) volume. (2) A coarse-to-fine framework contains two projection networks, providing more accurate localization of the target. (3) Based on the coarse-to-fine model, we transfer the organ boundingbox (B-box) detection to locating six extreme points along x, y and z directions in the query volume. Experiments on multi-organ localization from head-and-neck (HaN) CT volumes showed that our method acquired competitive performance in real time, which is more accurate and 10^5 times faster than template matching methods with the same setting. Code is available: https://github.com/LWHYC/RPR-Loc.
In this paper we consider a continuous description based on stochastic differential equations of the popular particle swarm optimization (PSO) process for solving global optimization problems and derive in the large particle limit the corresponding mean-field approximation based on Vlasov-Fokker-Planck-type equations. The disadvantage of memory effects induced by the need to store the local best position is overcome by the introduction of an additional differential equation describing the evolution of the local best. A regularization process for the global best permits to formally derive the respective mean-field description. Subsequently, in the small inertia limit, we compute the related macroscopic hydrodynamic equations that clarify the link with the recently introduced consensus based optimization (CBO) methods. Several numerical examples illustrate the mean field process, the small inertia limit and the potential of this general class of global optimization methods.
Hateful memes are widespread in social media and convey negative information. The main challenge of hateful memes detection is that the expressive meaning can not be well recognized by a single modality. In order to further integrate modal information, we investigate a candidate solution based on complementary visual and linguistic network in Hateful Memes Challenge 2020. In this way, more comprehensive information of the multi-modality could be explored in detail. Both contextual-level and sensitive object-level information are considered in visual and linguistic embedding to formulate the complex multi-modal scenarios. Specifically, a pre-trained classifier and object detector are utilized to obtain the contextual features and region-of-interests (RoIs) from the input, followed by the position representation fusion for visual embedding. While linguistic embedding is composed of three components, i.e., the sentence words embedding, position embedding and the corresponding Spacy embedding (Sembedding), which is a symbol represented by vocabulary extracted by Spacy. Both visual and linguistic embedding are fed into the designed Complementary Visual and Linguistic (CVL) networks to produce the prediction for hateful memes. Experimental results on Hateful Memes Challenge Dataset demonstrate that CVL provides a decent performance, and produces 78:48% and 72:95% on the criteria of AUROC and Accuracy. Code is available at https://github.com/webYFDT/hateful.
Arguably the key reason for the success of deep neural networks is their ability to autonomously form non-linear combinations of the input features, which can be used in subsequent layers of the network. The analogon to this capability in inductive rule learning is to learn a structured rule base, where the inputs are combined to learn new auxiliary concepts, which can then be used as inputs by subsequent rules. Yet, research on rule learning algorithms that have such capabilities is still in their infancy, which is - we would argue - one of the key impediments to substantial progress in this field. In this position paper, we want to draw attention to this unsolved problem, with a particular focus on previous work in predicate invention and multi-label rule learning