Paper Group NAWR 18
IL2M: Class Incremental Learning With Dual Memory. On the Summarization of Consumer Health Questions. Coresets for Archetypal Analysis. PointWeb: Enhancing Local Neighborhood Features for Point Cloud Processing. Embedding Strategies for Specialized Domains: Application to Clinical Entity Recognition. Measuring Gender Bias in Word Embeddings across …
IL2M: Class Incremental Learning With Dual Memory
| Title | IL2M: Class Incremental Learning With Dual Memory |
| Authors | Eden Belouadah, Adrian Popescu |
| Abstract | This paper presents a class incremental learning (IL) method which exploits fine tuning and a dual memory to reduce the negative effect of catastrophic forgetting in image recognition. First, we simplify the current fine tuning based approaches which use a combination of classification and distillation losses to compensate for the limited availability of past data. We find that the distillation term actually hurts performance when a memory is allowed. Then, we modify the usual class IL memory component. Similar to existing works, a first memory stores exemplar images of past classes. A second memory is introduced here to store past class statistics obtained when they were initially learned. The intuition here is that classes are best modeled when all their data are available and that their initial statistics are useful across different incremental states. A prediction bias towards newly learned classes appears during inference because the dataset is imbalanced in their favor. The challenge is to make predictions of new and past classes more comparable. To do this, scores of past classes are rectified by leveraging contents from both memories. The method has negligible added cost, both in terms of memory and of inference complexity. Experiments with three large public datasets show that the proposed approach is more effective than a range of competitive state-of-the-art methods. |
| Tasks | |
| Published | 2019-10-01 |
| URL | http://openaccess.thecvf.com/content_ICCV_2019/html/Belouadah_IL2M_Class_Incremental_Learning_With_Dual_Memory_ICCV_2019_paper.html |
| http://openaccess.thecvf.com/content_ICCV_2019/papers/Belouadah_IL2M_Class_Incremental_Learning_With_Dual_Memory_ICCV_2019_paper.pdf | |
| PWC | https://paperswithcode.com/paper/il2m-class-incremental-learning-with-dual |
| Repo | https://github.com/EdenBelouadah/class-incremental-learning |
| Framework | pytorch |
On the Summarization of Consumer Health Questions
| Title | On the Summarization of Consumer Health Questions |
| Authors | Asma Ben Abacha, Dina Demner-Fushman |
| Abstract | Question understanding is one of the main challenges in question answering. In real world applications, users often submit natural language questions that are longer than needed and include peripheral information that increases the complexity of the question, leading to substantially more false positives in answer retrieval. In this paper, we study neural abstractive models for medical question summarization. We introduce the MeQSum corpus of 1,000 summarized consumer health questions. We explore data augmentation methods and evaluate state-of-the-art neural abstractive models on this new task. In particular, we show that semantic augmentation from question datasets improves the overall performance, and that pointer-generator networks outperform sequence-to-sequence attentional models on this task, with a ROUGE-1 score of 44.16{%}. We also present a detailed error analysis and discuss directions for improvement that are specific to question summarization. |
| Tasks | Data Augmentation, Question Answering |
| Published | 2019-07-01 |
| URL | https://www.aclweb.org/anthology/P19-1215/ |
| https://www.aclweb.org/anthology/P19-1215 | |
| PWC | https://paperswithcode.com/paper/on-the-summarization-of-consumer-health |
| Repo | https://github.com/abachaa/MeQSum |
| Framework | none |
Coresets for Archetypal Analysis
| Title | Coresets for Archetypal Analysis |
| Authors | Sebastian Mair, Ulf Brefeld |
| Abstract | Archetypal analysis represents instances as linear mixtures of prototypes (the archetypes) that lie on the boundary of the convex hull of the data. Archetypes are thus often better interpretable than factors computed by other matrix factorization techniques. However, the interpretability comes with high computational cost due to additional convexity-preserving constraints. In this paper, we propose efficient coresets for archetypal analysis. Theoretical guarantees are derived by showing that quantization errors of k-means upper bound archetypal analysis; the computation of a provable absolute-coreset can be performed in only two passes over the data. Empirically, we show that the coresets lead to improved performance on several data sets. |
| Tasks | Quantization |
| Published | 2019-12-01 |
| URL | http://papers.nips.cc/paper/8945-coresets-for-archetypal-analysis |
| http://papers.nips.cc/paper/8945-coresets-for-archetypal-analysis.pdf | |
| PWC | https://paperswithcode.com/paper/coresets-for-archetypal-analysis |
| Repo | https://github.com/smair/archetypalanalysis-coreset |
| Framework | none |
PointWeb: Enhancing Local Neighborhood Features for Point Cloud Processing
| Title | PointWeb: Enhancing Local Neighborhood Features for Point Cloud Processing |
| Authors | Hengshuang Zhao, Li Jiang, Chi-Wing Fu, Jiaya Jia |
| Abstract | This paper presents PointWeb, a new approach to extract contextual features from local neighborhood in a point cloud. Unlike previous work, we densely connect each point with every other in a local neighborhood, aiming to specify feature of each point based on the local region characteristics for better representing the region. A novel module, namely Adaptive Feature Adjustment (AFA) module, is presented to find the interaction between points. For each local region, an impact map carrying element-wise impact between point pairs is applied to the feature difference map. Each feature is then pulled or pushed by other features in the same region according to the adaptively learned impact indicators. The adjusted features are well encoded with region information, and thus benefit the point cloud recognition tasks, such as point cloud segmentation and classification. Experimental results show that our model outperforms the state-of-the-arts on both semantic segmentation and shape classification datasets. |
| Tasks | Semantic Segmentation |
| Published | 2019-06-01 |
| URL | http://openaccess.thecvf.com/content_CVPR_2019/html/Zhao_PointWeb_Enhancing_Local_Neighborhood_Features_for_Point_Cloud_Processing_CVPR_2019_paper.html |
| http://openaccess.thecvf.com/content_CVPR_2019/papers/Zhao_PointWeb_Enhancing_Local_Neighborhood_Features_for_Point_Cloud_Processing_CVPR_2019_paper.pdf | |
| PWC | https://paperswithcode.com/paper/pointweb-enhancing-local-neighborhood |
| Repo | https://github.com/hszhao/PointWeb |
| Framework | pytorch |
Embedding Strategies for Specialized Domains: Application to Clinical Entity Recognition
| Title | Embedding Strategies for Specialized Domains: Application to Clinical Entity Recognition |
| Authors | Hicham El Boukkouri, Olivier Ferret, Thomas Lavergne, Pierre Zweigenbaum |
| Abstract | Using pre-trained word embeddings in conjunction with Deep Learning models has become the {``}de facto{''} approach in Natural Language Processing (NLP). While this usually yields satisfactory results, off-the-shelf word embeddings tend to perform poorly on texts from specialized domains such as clinical reports. Moreover, training specialized word representations from scratch is often either impossible or ineffective due to the lack of large enough in-domain data. In this work, we focus on the clinical domain for which we study embedding strategies that rely on general-domain resources only. We show that by combining off-the-shelf contextual embeddings (ELMo) with static word2vec embeddings trained on a small in-domain corpus built from the task data, we manage to reach and sometimes outperform representations learned from a large corpus in the medical domain. | |
| Tasks | Clinical Concept Extraction, Word Embeddings |
| Published | 2019-07-01 |
| URL | https://www.aclweb.org/anthology/P19-2041/ |
| https://www.aclweb.org/anthology/P19-2041 | |
| PWC | https://paperswithcode.com/paper/embedding-strategies-for-specialized-domains |
| Repo | https://github.com/helboukkouri/acl_srw_2019 |
| Framework | pytorch |
Measuring Gender Bias in Word Embeddings across Domains and Discovering New Gender Bias Word Categories
| Title | Measuring Gender Bias in Word Embeddings across Domains and Discovering New Gender Bias Word Categories |
| Authors | Kaytlin Chaloner, Alfredo Maldonado |
| Abstract | Prior work has shown that word embeddings capture human stereotypes, including gender bias. However, there is a lack of studies testing the presence of specific gender bias categories in word embeddings across diverse domains. This paper aims to fill this gap by applying the WEAT bias detection method to four sets of word embeddings trained on corpora from four different domains: news, social networking, biomedical and a gender-balanced corpus extracted from Wikipedia (GAP). We find that some domains are definitely more prone to gender bias than others, and that the categories of gender bias present also vary for each set of word embeddings. We detect some gender bias in GAP. We also propose a simple but novel method for discovering new bias categories by clustering word embeddings. We validate this method through WEAT{'}s hypothesis testing mechanism and find it useful for expanding the relatively small set of well-known gender bias word categories commonly used in the literature. |
| Tasks | Gender Bias Detection, Word Embeddings |
| Published | 2019-08-01 |
| URL | https://www.aclweb.org/anthology/W19-3804/ |
| https://www.aclweb.org/anthology/W19-3804 | |
| PWC | https://paperswithcode.com/paper/measuring-gender-bias-in-word-embeddings |
| Repo | https://github.com/alfredomg/GeBNLP2019 |
| Framework | none |
Learning to Abstract for Memory-augmented Conversational Response Generation
| Title | Learning to Abstract for Memory-augmented Conversational Response Generation |
| Authors | Zhiliang Tian, Wei Bi, Xiaopeng Li, Nevin L. Zhang |
| Abstract | Neural generative models for open-domain chit-chat conversations have become an active area of research in recent years. A critical issue with most existing generative models is that the generated responses lack informativeness and diversity. A few researchers attempt to leverage the results of retrieval models to strengthen the generative models, but these models are limited by the quality of the retrieval results. In this work, we propose a memory-augmented generative model, which learns to abstract from the training corpus and saves the useful information to the memory to assist the response generation. Our model clusters query-response samples, extracts characteristics of each cluster, and learns to utilize these characteristics for response generation. Experimental results show that our model outperforms other competitive baselines. |
| Tasks | Conversational Response Generation |
| Published | 2019-07-01 |
| URL | https://www.aclweb.org/anthology/P19-1371/ |
| https://www.aclweb.org/anthology/P19-1371 | |
| PWC | https://paperswithcode.com/paper/learning-to-abstract-for-memory-augmented |
| Repo | https://github.com/tianzhiliang/MemoryAugDialog |
| Framework | pytorch |
Towards Multimodal Sarcasm Detection (An _Obviously_ Perfect Paper)
| Title | Towards Multimodal Sarcasm Detection (An _Obviously_ Perfect Paper) |
| Authors | Santiago Castro, Devamanyu Hazarika, Ver{'o}nica P{'e}rez-Rosas, Roger Zimmermann, Rada Mihalcea, Soujanya Poria |
| Abstract | Sarcasm is often expressed through several verbal and non-verbal cues, e.g., a change of tone, overemphasis in a word, a drawn-out syllable, or a straight looking face. Most of the recent work in sarcasm detection has been carried out on textual data. In this paper, we argue that incorporating multimodal cues can improve the automatic classification of sarcasm. As a first step towards enabling the development of multimodal approaches for sarcasm detection, we propose a new sarcasm dataset, Multimodal Sarcasm Detection Dataset (MUStARD), compiled from popular TV shows. MUStARD consists of audiovisual utterances annotated with sarcasm labels. Each utterance is accompanied by its context of historical utterances in the dialogue, which provides additional information on the scenario where the utterance occurs. Our initial results show that the use of multimodal information can reduce the relative error rate of sarcasm detection by up to 12.9{%} in F-score when compared to the use of individual modalities. The full dataset is publicly available for use at https://github.com/soujanyaporia/MUStARD. |
| Tasks | Sarcasm Detection |
| Published | 2019-07-01 |
| URL | https://www.aclweb.org/anthology/P19-1455/ |
| https://www.aclweb.org/anthology/P19-1455 | |
| PWC | https://paperswithcode.com/paper/towards-multimodal-sarcasm-detection-an-1 |
| Repo | https://github.com/soujanyaporia/MUStARD |
| Framework | pytorch |
FASTDial: Abstracting Dialogue Policies for Fast Development of Task Oriented Agents
| Title | FASTDial: Abstracting Dialogue Policies for Fast Development of Task Oriented Agents |
| Authors | Serra Sinem Tekiroglu, Bernardo Magnini, Marco Guerini |
| Abstract | We present a novel abstraction framework called FASTDial for designing task oriented dialogue agents, built on top of the OpenDial toolkit. This framework is meant to facilitate prototyping and development of dialogue systems from scratch also by non tech savvy especially when limited training data is available. To this end, we use a generic and simple frame-slots data-structure with pre-defined dialogue policies that allows for fast design and implementation at the price of some flexibility reduction. Moreover, it allows for minimizing programming effort and domain expert training time, by hiding away many implementation details. We provide a system demonstration screencast video in the following link: https://vimeo.com/329840716 |
| Tasks | |
| Published | 2019-07-01 |
| URL | https://www.aclweb.org/anthology/P19-3013/ |
| https://www.aclweb.org/anthology/P19-3013 | |
| PWC | https://paperswithcode.com/paper/fastdial-abstracting-dialogue-policies-for |
| Repo | https://github.com/serrasinem/FASTDial |
| Framework | none |
A Creative Dialog Generator for Fallout 4
| Title | A Creative Dialog Generator for Fallout 4 |
| Authors | Khalid Alnajjar, Mika Hämäläinen |
| Abstract | This software demonstration describes a mod for Fallout 4 that will adapt in-game dialog to the context of the current state of the game. The dialog is generated by a computationally creative back-end software during the game play. The mod solves the problem of Fallout 4 not supporting dynamically generated dialog by showing dialog in an overlay application on top of the game window |
| Tasks | |
| Published | 2019-08-01 |
| URL | https://dl.acm.org/doi/10.1145/3337722.3341865 |
| https://helda.helsinki.fi//bitstream/handle/10138/307172/Alnajjar_paper_176.pdf?sequence=1 | |
| PWC | https://paperswithcode.com/paper/a-creative-dialog-generator-for-fallout-4 |
| Repo | https://github.com/mikahama/fallout4-dynamic-dialog |
| Framework | none |
Creative Contextual Dialog Adaptation in an Open World RPG
| Title | Creative Contextual Dialog Adaptation in an Open World RPG |
| Authors | Mika Hämäläinen, Khalid Alnajjar |
| Abstract | Role playing games rely typically on hand-written dialog that has no flexibility in adapting to the game state such as the level of the player. This is an even bigger problem for open world RPGs that make it possible to complete the game quests and objectives virtually in any given order. We present a computationally creative method for adapting Fallout 4 dialog to the changes in the game state using word embeddings for semantics and a BRNN for sequence-to-sequence paraphrasing of syntax. |
| Tasks | Word Embeddings |
| Published | 2019-08-01 |
| URL | https://dl.acm.org/doi/10.1145/3337722.3341824 |
| https://helda.helsinki.fi//bitstream/handle/10138/307173/W6_01_H_m_l_inen.pdf?sequence=1 | |
| PWC | https://paperswithcode.com/paper/creative-contextual-dialog-adaptation-in-an |
| Repo | https://github.com/mikahama/fallout4-dynamic-dialog |
| Framework | none |
3D color homography model for photo-realistic color transfer re-coding
| Title | 3D color homography model for photo-realistic color transfer re-coding |
| Authors | Han Gong; Graham Finlayson; Robert Fisher; Fufu Fang |
| Abstract | Color transfer is an image editing process that naturally transfers the color theme of a source image to a target image. In this paper, we propose a 3D color homography model which approximates photo-realistic color transfer algorithm as a combination of a 3D perspective transform and a mean intensity mapping. A key advantage of our approach is that the re-coded color transfer algorithm is simple and accurate. Our evaluation demonstrates that our 3D color homography model delivers leading color transfer re-coding performance. In addition, we also show that our 3D color homography model can be applied to color transfer artifact fixing, complex color transfer acceleration, and color-robust image stitching. |
| Tasks | Image Stitching |
| Published | 2019-03-01 |
| URL | https://link.springer.com/article/10.1007/s00371-017-1462-x |
| https://link.springer.com/article/10.1007/s00371-017-1462-x | |
| PWC | https://paperswithcode.com/paper/3d-color-homography-model-for-photo-realistic |
| Repo | https://github.com/hangong/h3d-cf |
| Framework | none |
Experiment-based detection of service disruption attacks in optical networks using data analytics and unsupervised learning
| Title | Experiment-based detection of service disruption attacks in optical networks using data analytics and unsupervised learning |
| Authors | Marija Furdek, Carlos Natalino, Marco Schiano and Andrea Di Giglio |
| Abstract | The paper addresses the detection of malicious attacks targeting service disruption at the optical layer as a key prerequisite for fast and effective attack response and network recovery. We experimentally demonstrate the effects of signal insertion attacks with varying intensity in a real-life scenario. By applying data analytics tools, we analyze the properties of the obtained dataset to determine how the relationships among different optical performance monitoring (OPM) parameters of the signal change in the presence of an attack as opposed to the normal operating conditions. In addition, we evaluate the performance of an unsupervised learning technique, i.e., a clustering algorithm for anomaly detection, which can detect attacks as anomalies without prior knowledge of the attacks. We demonstrate the potential and the challenges of unsupervised learning for attack detection, propose guidelines for attack signature identification needed for the detection of the considered attack methods, and discuss remaining challenges related to optical network security. |
| Tasks | Anomaly Detection |
| Published | 2019-02-01 |
| URL | https://www.researchgate.net/publication/330799002_Experiment-based_detection_of_service_disruption_attacks_in_optical_networks_using_data_analytics_and_unsupervised_learning |
| https://www.researchgate.net/publication/330799002_Experiment-based_detection_of_service_disruption_attacks_in_optical_networks_using_data_analytics_and_unsupervised_learning | |
| PWC | https://paperswithcode.com/paper/experiment-based-detection-of-service |
| Repo | https://github.com/carlosnatalino/2019-spie-photonics-west |
| Framework | none |
Gaussian-Based Pooling for Convolutional Neural Networks
| Title | Gaussian-Based Pooling for Convolutional Neural Networks |
| Authors | Takumi Kobayashi |
| Abstract | Convolutional neural networks (CNNs) contain local pooling to effectively downsize feature maps for increasing computation efficiency as well as robustness to input variations. The local pooling methods are generally formulated in a form of convex combination of local neuron activations for retaining the characteristics of an input feature map in a manner similar to image downscaling. In this paper, to improve performance of CNNs, we propose a novel local pooling method based on the Gaussian-based probabilistic model over local neuron activations for flexibly pooling (extracting) features, in contrast to the previous model restricting the output within the convex hull of local neurons. In the proposed method, the local neuron activations are aggregated into the statistics of mean and standard deviation in a Gaussian distribution, and then on the basis of those statistics, we construct the probabilistic model suitable for the pooling in accordance with the knowledge about local pooling in CNNs. Through the probabilistic model equipped with trainable parameters, the proposed method naturally integrates two schemes of adaptively training the pooling form based on input feature maps and stochastically performing the pooling throughout the end-to-end learning. The experimental results on image classification demonstrate that the proposed method favorably improves performance of various CNNs in comparison with the other pooling methods. |
| Tasks | Image Classification |
| Published | 2019-12-01 |
| URL | http://papers.nips.cc/paper/9300-gaussian-based-pooling-for-convolutional-neural-networks |
| http://papers.nips.cc/paper/9300-gaussian-based-pooling-for-convolutional-neural-networks.pdf | |
| PWC | https://paperswithcode.com/paper/gaussian-based-pooling-for-convolutional |
| Repo | https://github.com/tk1980/GaussianPooling |
| Framework | pytorch |
Overcoming the Disentanglement vs Reconstruction Trade-off via Jacobian Supervision
| Title | Overcoming the Disentanglement vs Reconstruction Trade-off via Jacobian Supervision |
| Authors | José Lezama |
| Abstract | A major challenge in learning image representations is the disentangling of the factors of variation underlying the image formation. This is typically achieved with an autoencoder architecture where a subset of the latent variables is constrained to correspond to specific factors, and the rest of them are considered nuisance variables. This approach has an important drawback: as the dimension of the nuisance variables is increased, image reconstruction is improved, but the decoder has the flexibility to ignore the specified factors, thus losing the ability to condition the output on them. In this work, we propose to overcome this trade-off by progressively growing the dimension of the latent code, while constraining the Jacobian of the output image with respect to the disentangled variables to remain the same. As a result, the obtained models are effective at both disentangling and reconstruction. We demonstrate the applicability of this method in both unsupervised and supervised scenarios for learning disentangled representations. In a facial attribute manipulation task, we obtain high quality image generation while smoothly controlling dozens of attributes with a single model. This is an order of magnitude more disentangled factors than state-of-the-art methods, while obtaining visually similar or superior results, and avoiding adversarial training. |
| Tasks | Image Generation, Image Reconstruction |
| Published | 2019-05-01 |
| URL | https://openreview.net/forum?id=Hkg4W2AcFm |
| https://openreview.net/pdf?id=Hkg4W2AcFm | |
| PWC | https://paperswithcode.com/paper/overcoming-the-disentanglement-vs |
| Repo | https://github.com/jlezama/disentangling-jacobian |
| Framework | pytorch |
