ArXiv Weekly Radiostation：本週NLP、CV、ML精選論文30篇（1.5-1.12）

Multiplex Word Embeddings for Selectional Preference Acquisition. 

Conventional word embeddings represent words with fixed vectors, which are usually trained based on co-occurrence patterns among words. In doing so, however, the power of such representations is limited, where the same word might be functionalized separately under different syntactic relations. To address this limitation, one solution is to incorporate relational dependencies of different words into their embeddings. Therefore, in this paper, we propose a multiplex word embedding model, which can be easily extended according to various relations among words. As a result, each word has a center embedding to represent its overall semantics, and several relational embeddings to represent its relational dependencies. Compared to existing models, our model can effectively distinguish words with respect to different relations without introducing unnecessary sparseness. Moreover, to accommodate various relations, we use a small dimension for relational embeddings and our model is able to keep their effectiveness. Experiments on selectional preference acquisition and word similarity demonstrate the effectiveness of the proposed model, and a further study of scalability also proves that our embeddings only need 1/20 of the original embedding size to achieve better performance.

02

REST: A thread embedding approach for identifying and classifying user-specified information in security forums. 

How can we extract useful information from a security forum? We focus on identifying threads of interest to a security professional: (a) alerts of worrisome events, such as attacks, (b) offering of malicious services and products, (c) hacking information to perform malicious acts, and (d) useful security-related experiences. The analysis of security forums is in its infancy despite several promising recent works. Novel approaches are needed to address the challenges in this domain: (a) the difficulty in specifying the "topics" of interest efficiently, and (b) the unstructured and informal nature of the text. We propose, REST, a systematic methodology to: (a) identify threads of interest based on a, possibly incomplete, bag of words, and (b) classify them into one of the four classes above. The key novelty of the work is a multi-step weighted embedding approach: we project words, threads and classes in appropriate embedding spaces and establish relevance and similarity there. We evaluate our method with real data from three security forums with a total of 164k posts and 21K threads. First, REST robustness to initial keyword selection can extend the user-provided keyword set and thus, it can recover from missing keywords. Second, REST categorizes the threads into the classes of interest with superior accuracy compared to five other methods: REST exhibits an accuracy between 63.3-76.9%. We see our approach as a first step for harnessing the wealth of information of online forums in a user-friendly way, since the user can loosely specify her keywords of interest.

03

A Neural Approach to Discourse Relation Signal Detection. 

Previous data-driven work investigating the types and distributions of discourse relation signals, including discourse markers such as 'however' or phrases such as 'as a result' has focused on the relative frequencies of signal words within and outside text from each discourse relation. Such approaches do not allow us to quantify the signaling strength of individual instances of a signal on a scale (e.g. more or less discourse-relevant instances of 'and'), to assess the distribution of ambiguity for signals, or to identify words that hinder discourse relation identification in context ('anti-signals' or 'distractors'). In this paper we present a data-driven approach to signal detection using a distantly supervised neural network and develop a metric, {\Delta}s (or 'delta-softmax'), to quantify signaling strength. Ranging between -1 and 1 and relying on recent advances in contextualized words embeddings, the metric represents each word's positive or negative contribution to the identifiability of a relation in specific instances in context. Based on an English corpus annotated for discourse relations using Rhetorical Structure Theory and signal type annotations anchored to specific tokens, our analysis examines the reliability of the metric, the places where it overlaps with and differs from human judgments, and the implications for identifying features that neural models may need in order to perform better on automatic discourse relation classification.

04

Improving Entity Linking by Modeling Latent Entity Type Information. 

Existing state of the art neural entity linking models employ attention-based bag-of-words context model and pre-trained entity embeddings bootstrapped from word embeddings to assess topic level context compatibility. However, the latent entity type information in the immediate context of the mention is neglected, which causes the models often link mentions to incorrect entities with incorrect type. To tackle this problem, we propose to inject latent entity type information into the entity embeddings based on pre-trained BERT. In addition, we integrate a BERT-based entity similarity score into the local context model of a state-of-the-art model to better capture latent entity type information. Our model significantly outperforms the state-of-the-art entity linking models on standard benchmark (AIDA-CoNLL). Detailed experiment analysis demonstrates that our model corrects most of the type errors produced by the direct baseline.

Learning Bilingual Word Embeddings Using Lexical Definitions. 

Bilingual word embeddings, which representlexicons of different languages in a shared em-bedding space, are essential for supporting se-mantic and knowledge transfers in a variety ofcross-lingual NLP tasks. Existing approachesto training bilingual word embeddings requireoften require pre-defined seed lexicons that areexpensive to obtain, or parallel sentences thatcomprise coarse and noisy alignment. In con-trast, we propose BilLex that leverages pub-licly available lexical definitions for bilingualword embedding learning. Without the needof predefined seed lexicons, BilLex comprisesa novel word pairing strategy to automati-cally identify and propagate the precise fine-grained word alignment from lexical defini-tions. We evaluate BilLex in word-level andsentence-level translation tasks, which seek tofind the cross-lingual counterparts of wordsand sentences respectively.BilLex signifi-cantly outperforms previous embedding meth-ods on both tasks.

06

Stance Detection Benchmark: How Robust Is Your Stance Detection? 

Stance Detection (StD) aims to detect an author's stance towards a certain topic or claim and has become a key component in applications like fake news detection, claim validation, and argument search. However, while stance is easily detected by humans, machine learning models are clearly falling short of this task. Given the major differences in dataset sizes and framing of StD (e.g. number of classes and inputs), we introduce a StD benchmark that learns from ten StD datasets of various domains in a multi-dataset learning (MDL) setting, as well as from related tasks via transfer learning. Within this benchmark setup, we are able to present new state-of-the-art results on five of the datasets. Yet, the models still perform well below human capabilities and even simple adversarial attacks severely hurt the performance of MDL models. Deeper investigation into this phenomenon suggests the existence of biases inherited from multiple datasets by design. Our analysis emphasizes the need of focus on robustness and de-biasing strategies in multi-task learning approaches. The benchmark dataset and code is made available.

07

Leveraging Prior Knowledge for Protein-Protein Interaction Extraction with Memory Network. 

Automatically extracting Protein-Protein Interactions (PPI) from biomedical literature provides additional support for precision medicine efforts. This paper proposes a novel memory network-based model (MNM) for PPI extraction, which leverages prior knowledge about protein-protein pairs with memory networks. The proposed MNM captures important context clues related to knowledge representations learned from knowledge bases. Both entity embeddings and relation embeddings of prior knowledge are effective in improving the PPI extraction model, leading to a new state-of-the-art performance on the BioCreative VI PPI dataset. The paper also shows that multiple computational layers over an external memory are superior to long short-term memory networks with the local memories.

The correct use of Dutch pronouns 'die' and 'dat' is a stumbling block for both native and non-native speakers of Dutch due to the multiplicity of syntactic functions and the dependency on the antecedent's gender and number. Drawing on previous research conducted on neural context-dependent dt-mistake correction models (Heyman et al. 2018), this study constructs the first neural network model for Dutch demonstrative and relative pronoun resolution that specifically focuses on the correction and part-of-speech prediction of these two pronouns. Two separate datasets are built with sentences obtained from, respectively, the Dutch Europarl corpus (Koehn 2015) - which contains the proceedings of the European Parliament from 1996 to the present - and the SoNaR corpus (Oostdijk et al. 2013) - which contains Dutch texts from a variety of domains such as newspapers, blogs and legal texts. Firstly, a binary classification model solely predicts the correct 'die' or 'dat'. The classifier with a bidirectional long short-term memory architecture achieves 84.56% accuracy. Secondly, a multitask classification model simultaneously predicts the correct 'die' or 'dat' and its part-of-speech tag. The model containing a combination of a sentence and context encoder with both a bidirectional long short-term memory architecture results in 88.63% accuracy for die/dat prediction and 87.73% accuracy for part-of-speech prediction. More evenly-balanced data, larger word embeddings, an extra bidirectional long short-term memory layer and integrated part-of-speech knowledge positively affects die/dat prediction performance, while a context encoder architecture raises part-of-speech prediction performance. This study shows promising results and can serve as a starting point for future research on machine learning models for Dutch anaphora resolution.

Paraphrase Generation with Latent Bag of Words. 

Paraphrase generation is a longstanding important problem in natural language processing.In addition, recent progress in deep generative models has shown promising results on discrete latent variables for text generation.

Inspired by variational autoencoders with discrete latent structures, in this work, we propose a latent bag of words (BOW) model for paraphrase generation.We ground the semantics of a discrete latent variable by the BOW from the target sentences.We use this latent variable to build a fully differentiable content planning and surface realization model.Specifically, we use source words to predict their neighbors and model the target BOW with a mixture of softmax.We use Gumbel top-k reparameterization to perform differentiable subset sampling from the predicted BOW distribution.We retrieve the sampled word embeddings and use them to augment the decoder and guide its generation search space.

Our latent BOW model not only enhances the decoder, but also exhibits clear interpretability.

We show the model interpretability with regard to \emph{(i)} unsupervised learning of word neighbors \emph{(ii)} the step-by-step generation procedure.

Extensive experiments demonstrate the transparent and effective generation process of this model.\footnote{Our code can be found at \url{https://github.com/FranxYao/dgm_latent_bow}}

Adapting Deep Learning for Sentiment Classification of Code-Switched Informal Short Text. 

Agriculture-Vision: A Large Aerial Image Database for Agricultural Pattern Analysis. 

Weakly Supervised Visual Semantic Parsing. 

Bridging Knowledge Graphs to Generate Scene Graphs. 

Scene graphs are powerful representations that encode images into their abstract semantic elements, i.e, objects and their interactions, which facilitates visual comprehension and explainable reasoning. On the other hand, commonsense knowledge graphs are rich repositories that encode how the world is structured, and how general concepts interact. In this paper, we present a unified formulation of these two constructs, where a scene graph is seen as an image-conditioned instantiation of a commonsense knowledge graph. Based on this new perspective, we re-formulate scene graph generation as the inference of a bridge between the scene and commonsense graphs, where each entity or predicate instance in the scene graph has to be linked to its corresponding entity or predicate class in the commonsense graph. To this end, we propose a heterogeneous graph inference framework allowing to exploit the rich structure within the scene and commonsense at the same time. Through extensive experiments, we show the proposed method achieves significant improvement over the state of the art.

General Partial Label Learning via Dual Bipartite Graph Autoencoder. 

RPR: Random Partition Relaxation for Training; Binary and Ternary Weight Neural Networks. 

We present Random Partition Relaxation (RPR), a method for strong quantization of neural networks weight to binary (+1/-1) and ternary (+1/0/-1) values. Starting from a pre-trained model, we quantize the weights and then relax random partitions of them to their continuous values for retraining before re-quantizing them and switching to another weight partition for further adaptation. We demonstrate binary and ternary-weight networks with accuracies beyond the state-of-the-art for GoogLeNet and competitive performance for ResNet-18 and ResNet-50 using an SGD-based training method that can easily be integrated into existing frameworks.

06

Chained Representation Cycling: Learning to Estimate 3D Human Pose and Shape by Cycling Between Representations. 

The goal of many computer vision systems is to transform image pixels into 3D representations. Recent popular models use neural networks to regress directly from pixels to 3D object parameters. Such an approach works well when supervision is available, but in problems like human pose and shape estimation, it is difficult to obtain natural images with 3D ground truth. To go one step further, we propose a new architecture that facilitates unsupervised, or lightly supervised, learning. The idea is to break the problem into a series of transformations between increasingly abstract representations. Each step involves a cycle designed to be learnable without annotated training data, and the chain of cycles delivers the final solution. Specifically, we use 2D body part segments as an intermediate representation that contains enough information to be lifted to 3D, and at the same time is simple enough to be learned in an unsupervised way. We demonstrate the method by learning 3D human pose and shape from un-paired and un-annotated images. We also explore varying amounts of paired data and show that cycling greatly alleviates the need for paired data. While we present results for modeling humans, our formulation is general and can be applied to other vision problems.

Few-shot Learning with Multi-scale Self-supervision. 

Compression of convolutional neural networks for high performance imagematching tasks on mobile devices. 

Deep neural networks have demonstrated state-of-the-art performance for feature-based image matching through the advent of new large and diverse datasets. However, there has been little work on evaluating the computational cost, model size, and matching accuracy tradeoffs for these models. This paper explicitly addresses these practical constraints by considering state-of-the-art L2Net architecture. We observe a significant redundancy in the L2Net architecture, which we exploit through the use of depthwise separable layers and an efficient Tucker decomposition. We demonstrate that a combination of these methods is more effective, but still sacrifices the top-end accuracy. We therefore propose the Convolution-Depthwise-Pointwise (CDP) layer, which provides a means of interpolating between the standard and depthwise separable convolutions. With this proposed layer, we are able to achieve up to 8 times reduction in the number of parameters on the L2Net architecture, 13 times reduction in the computational complexity, while sacrificing less than 1% on the overall accuracy across the HPatches benchmarks. To further demonstrate the generalisation of this approach, we apply it to the SuperPoint model. We show that CDP layers improve upon the accuracy while using significantly less parameters and floating-point operations for inference.

MW-GAN: Multi-Warping GAN for Caricature Generation with Multi-Style Geometric Exaggeration. 

Given an input face photo, the goal of caricature generation is to produce stylized, exaggerated caricatures that share the same identity as the photo. It requires simultaneous style transfer and shape exaggeration with rich diversity, and meanwhile preserving the identity of the input. To address this challenging problem, we propose a novel framework called Multi-Warping GAN (MW-GAN), including a style network and a geometric network that are designed to conduct style transfer and geometric exaggeration respectively. We bridge the gap between the style and landmarks of an image with corresponding latent code spaces by a dual way design, so as to generate caricatures with arbitrary styles and geometric exaggeration, which can be specified either through random sampling of latent code or from a given caricature sample. Besides, we apply identity preserving loss to both image space and landmark space, leading to a great improvement in quality of generated caricatures. Experiments show that caricatures generated by MW-GAN have better quality than existing methods.

FrequentNet : A New Deep Learning Baseline for Image Classification. 

In this paper, we generalize the idea from the method called "PCANet" to achieve a new baseline deep learning model for image classification. Instead of using principal component vectors as the filter vector in "PCANet", we use basis vectors in discrete Fourier analysis and wavelets analysis as our filter vectors. Both of them achieve comparable performance to "PCANet" in benchmark datasets. It is noticeable that our algorithms do not require any optimization techniques to get those basis.

Why you may want to read this: Newest paper from Vipin Kumar (University of Minnesota).

Guruprasad Nayak, Rahul Ghosh, Xiaowei Jia, Varun Mithal, Vipin Kumar

02

Why you may want to read this: Newest paper from Ruslan Salakhutdinov (Associate Professor, Machine Learning Department, CMU), Louis-Philippe Morency (Associate professor,  Carnegie Mellon University).

Paul Pu Liang, Terrance Liu, Liu Ziyin, Ruslan Salakhutdinov, Louis-Philippe Morency

03

Why you may want to read this: Newest paper from Dieter Fox (Professor of Computer Science and Engineering, University of Washington, and Nvidia …).

Mohak Bhardwaj, Ankur Handa, Dieter Fox, Byron Boots

Model-free Reinforcement Learning (RL) algorithms work well in sequential decision-making problems when experience can be collected cheaply and model-based RL is effective when system dynamics can be modeled accurately. However, both of these assumptions can be violated in real world problems such as robotics, where querying the system can be prohibitively expensive and real-world dynamics can be difficult to model accurately. Although sim-to-real approaches such as domain randomization attempt to mitigate the effects of biased simulation,they can still suffer from optimization challenges such as local minima and hand-designed distributions for randomization, making it difficult to learn an accurate global value function or policy that directly transfers to the real world. In contrast to RL, Model Predictive Control (MPC) algorithms use a simulator to optimize a simple policy class online, constructing a closed-loop controller that can effectively contend with real-world dynamics. MPC performance is usually limited by factors such as model bias and the limited horizon of optimization. In this work, we present a novel theoretical connection between information theoretic MPC and entropy regularized RL and develop a Q-learning algorithm that can leverage biased models. We validate the proposed algorithm on sim-to-sim control tasks to demonstrate the improvements over optimal control and reinforcement learning from scratch. Our approach paves the way for deploying reinforcement learning algorithms on real-robots in a systematic manner.

04

Why you may want to read this: Newest paper from Daniel H. Mathalon (Professor of Psychiatry, University of California San Francisco), Judith M. Ford (UCSF, Stanford,Yale), Jessica A. Turner (Professor of Psychology and Neuroscience, Georgia State University), Vince D. Calhoun (Director-Translational Research in Neuroimaging and Data Science (TReNDS;GSU/GAtech …).

Haleh Falakshahi, Victor M. Vergara, Jingyu Liu, Daniel H. Mathalon, Judith M. Ford, James Voyvodic, Bryon A. Mueller, Aysenil Belger, Sarah McEwen, Steven G. Potkin, Adrian Preda, Hooman Rokham, Jing Sui, Jessica A. Turner, Sergey Plis, Vince D. Calhoun

Why you may want to read this: Newest paper from Ming Li (University Professor, University of Waterloo).

Shu-Ting Shi, Wenhao Zheng, Jun Tang, Qing-Guo Chen, Yao Hu, Jianke Zhu, Ming Li

06

Why you may want to read this: Newest paper from Ge Wang (Rensselaer Polytechnic Institute).

Why you may want to read this: Newest paper from Laurens van der Maaten (Research Scientist, Facebook AI Research), Kilian Q. Weinberger (Associate Professor of Computer Science, Cornell University, ASAPP Research).

Gao Huang, Zhuang Liu, Geoff Pleiss, Laurens van der Maaten, Kilian Q. Weinberger

08

Why you may want to read this: Newest paper from Chin-Teng Lin (University of Technology Sydney).

Zhenhua Shi, Dongrui Wu, Yu-Kai Wang, Chin-Teng Lin

Why you may want to read this: Newest paper from Chin-Teng Lin (University of Technology Sydney).

Yurui Ming, Dongrui Wu, Yu-Kai Wang, Yuhui Shi, Chin-Teng Lin

10

Why you may want to read this: Newest paper from Sebastien Ourselin (Professor of Healthcare Engineering, King's College London).

Lucas Fidon, Sebastien Ourselin, Tom Vercauteren