Paper Group NAWR 15
Google Vizier: A Service for Black-Box Optimization. Variational Bayesian Multiple Instance Learning With Gaussian Processes. A surprisingly effective out-of-the-box char2char model on the E2E NLG Challenge dataset. One Network to Solve Them All – Solving Linear Inverse Problems Using Deep Projection Models. Neural Disambiguation of Causal Lexical …
Google Vizier: A Service for Black-Box Optimization
| Title | Google Vizier: A Service for Black-Box Optimization |
| Authors | Daniel Golovin, Benjamin Solnik, Subhodeep Moitra, Greg Kochanski, John Karro, D. Sculley |
| Abstract | Any sufficiently complex system acts as a black box when it becomes easier to experiment with than to understand. Hence, black-box optimization has become increasingly important as systems have become more complex. In this paper we describe Google Vizier, a Google-internal service for performing black-box optimization that has become the de facto parameter tuning engine at Google. Google Vizier is used to optimize many of our machine learning models and other systems, and also provides core capabilities to Google’s Cloud Machine Learning HyperTune subsystem. We discuss our requirements, infrastructure design, underlying algorithms, and advanced features such as transfer learning and automated early stopping that the service provides. |
| Tasks | AutoML, Hyperparameter Optimization, Transfer Learning |
| Published | 2017-01-01 |
| URL | https://ai.google/research/pubs/pub46180 |
| https://ai.google/research/pubs/pub46180.pdf | |
| PWC | https://paperswithcode.com/paper/google-vizier-a-service-for-black-box |
| Repo | https://github.com/tobegit3hub/advisor |
| Framework | none |
Variational Bayesian Multiple Instance Learning With Gaussian Processes
| Title | Variational Bayesian Multiple Instance Learning With Gaussian Processes |
| Authors | Manuel Haussmann, Fred A. Hamprecht, Melih Kandemir |
| Abstract | Gaussian Processes (GPs) are effective Bayesian predictors. We here show for the first time that instance labels of a GP classifier can be inferred in the multiple instance learning (MIL) setting using variational Bayes. We achieve this via a new construction of the bag likelihood that assumes a large value if the instance predictions obey the MIL constraints and a small value otherwise. This construction lets us derive the update rules for the variational parameters analytically, assuring both scalable learning and fast convergence. We observe this model to improve the state of the art in instance label prediction from bag-level supervision in the 20 Newsgroups benchmark, as well as in Barrett’s cancer tumor localization from histopathology tissue microarray images. Furthermore, we introduce a novel pipeline for weakly supervised object detection naturally complemented with our model, which improves the state of the art on the PASCAL VOC 2007 and 2012 data sets. Last but not least, the performance of our model can be further boosted up using mixed supervision: a combination of weak (bag) and strong (instance) labels. |
| Tasks | Gaussian Processes, Multiple Instance Learning, Object Detection, Weakly Supervised Object Detection |
| Published | 2017-07-01 |
| URL | http://openaccess.thecvf.com/content_cvpr_2017/html/Haussmann_Variational_Bayesian_Multiple_CVPR_2017_paper.html |
| http://openaccess.thecvf.com/content_cvpr_2017/papers/Haussmann_Variational_Bayesian_Multiple_CVPR_2017_paper.pdf | |
| PWC | https://paperswithcode.com/paper/variational-bayesian-multiple-instance |
| Repo | https://github.com/manuelhaussmann/vgpmil |
| Framework | none |
A surprisingly effective out-of-the-box char2char model on the E2E NLG Challenge dataset
| Title | A surprisingly effective out-of-the-box char2char model on the E2E NLG Challenge dataset |
| Authors | Shubham Agarwal, Marc Dymetman |
| Abstract | We train a char2char model on the E2E NLG Challenge data, by exploiting {``}out-of-the-box{''} the recently released tfseq2seq framework, using some of the standard options offered by this tool. With minimal effort, and in particular without delexicalization, tokenization or lowercasing, the obtained raw predictions, according to a small scale human evaluation, are excellent on the linguistic side and quite reasonable on the adequacy side, the primary downside being the possible omissions of semantic material. However, in a significant number of cases (more than 70{%}), a perfect solution can be found in the top-20 predictions, indicating promising directions for solving the remaining issues. | |
| Tasks | Data-to-Text Generation, Text Generation, Tokenization |
| Published | 2017-08-01 |
| URL | https://www.aclweb.org/anthology/W17-5519/ |
| https://www.aclweb.org/anthology/W17-5519 | |
| PWC | https://paperswithcode.com/paper/a-surprisingly-effective-out-of-the-box |
| Repo | https://github.com/shubhamagarwal92/sigdialSubmission |
| Framework | tf |
One Network to Solve Them All – Solving Linear Inverse Problems Using Deep Projection Models
| Title | One Network to Solve Them All – Solving Linear Inverse Problems Using Deep Projection Models |
| Authors | J. H. Rick Chang, Chun-Liang Li, Barnabas Poczos, B. V. K. Vijaya Kumar, Aswin C. Sankaranarayanan |
| Abstract | While deep learning methods have achieved state-of-the-art performance in many challenging inverse problems like image inpainting and super-resolution, they invariably involve problem-specific training of the networks. Under this approach, each inverse problem requires its own dedicated network. In scenarios where we need to solve a wide variety of problems, e.g., on a mobile camera, it is inefficient and expensive to use these problem-specific networks. On the other hand, traditional methods using analytic signal priors can be used to solve any linear inverse problem; this often comes with a performance that is worse than learning-based methods. In this work, we provide a middle ground between the two kinds of methods – we propose a general framework to train a single deep neural network that solves arbitrary linear inverse problems. We achieve this by training a network that acts as a quasi-projection operator for the set of natural images and show that any linear inverse problem involving natural images can be solved using iterative methods. We empirically show that the proposed framework demonstrates superior performance over traditional methods using wavelet sparsity prior while achieving performance comparable to specially-trained networks on tasks including compressive sensing and pixel-wise inpainting. |
| Tasks | Compressive Sensing, Image Inpainting, Super-Resolution |
| Published | 2017-10-01 |
| URL | http://openaccess.thecvf.com/content_iccv_2017/html/Chang_One_Network_to_ICCV_2017_paper.html |
| http://openaccess.thecvf.com/content_ICCV_2017/papers/Chang_One_Network_to_ICCV_2017_paper.pdf | |
| PWC | https://paperswithcode.com/paper/one-network-to-solve-them-all-solving-linear-1 |
| Repo | https://github.com/image-science-lab/OneNet |
| Framework | tf |
Neural Disambiguation of Causal Lexical Markers Based on Context
| Title | Neural Disambiguation of Causal Lexical Markers Based on Context |
| Authors | Eugenio Mart{'\i}nez-C{'a}mara, Vered Shwartz, Iryna Gurevych, Ido Dagan |
| Abstract | |
| Tasks | Word Embeddings |
| Published | 2017-01-01 |
| URL | https://www.aclweb.org/anthology/W17-6927/ |
| https://www.aclweb.org/anthology/W17-6927 | |
| PWC | https://paperswithcode.com/paper/neural-disambiguation-of-causal-lexical |
| Repo | https://github.com/UKPLab/iwcs2017_disambiguation_causality_lexical_markers |
| Framework | tf |
Deep Co-Occurrence Feature Learning for Visual Object Recognition
| Title | Deep Co-Occurrence Feature Learning for Visual Object Recognition |
| Authors | Ya-Fang Shih, Yang-Ming Yeh, Yen-Yu Lin, Ming-Fang Weng, Yi-Chang Lu, Yung-Yu Chuang |
| Abstract | This paper addresses three issues in integrating part-based representations into convolutional neural networks (CNNs) for object recognition. First, most part-based models rely on a few pre-specified object parts. However, the optimal object parts for recognition often vary from category to category. Second, acquiring training data with part-level annotation is labor-intensive. Third, modeling spatial relationships between parts in CNNs often involves an exhaustive search of part templates over multiple network streams. We tackle the three issues by introducing a new network layer, called co-occurrence layer. It can extend a convolutional layer to encode the co-occurrence between the visual parts detected by the numerous neurons, instead of a few pre-specified parts. To this end, the feature maps serve as both filters and images, and mutual correlation filtering is conducted between them. The co-occurrence layer is end-to-end trainable. The resultant co-occurrence features are rotation- and translation-invariant, and are robust to object deformation. By applying this new layer to the VGG-16 and ResNet-152, we achieve the recognition rates of 83.6% and 85.8% on the Caltech-UCSD bird benchmark, respectively. The source code is available at https://github.com/yafangshih/Deep-COOC. |
| Tasks | Object Recognition |
| Published | 2017-07-01 |
| URL | http://openaccess.thecvf.com/content_cvpr_2017/html/Shih_Deep_Co-Occurrence_Feature_CVPR_2017_paper.html |
| http://openaccess.thecvf.com/content_cvpr_2017/papers/Shih_Deep_Co-Occurrence_Feature_CVPR_2017_paper.pdf | |
| PWC | https://paperswithcode.com/paper/deep-co-occurrence-feature-learning-for |
| Repo | https://github.com/yafangshih/Deep-COOC |
| Framework | none |
Co-trained convolutional neural networks for automated detection of prostate cancer in multi-parametric MRI
| Title | Co-trained convolutional neural networks for automated detection of prostate cancer in multi-parametric MRI |
| Authors | Xin Yang, Chaoyue Liu, Zhiwei Wang, Jun Yang, Hung Le Min, Liang Wang, Kwang-Ting (Tim) Cheng |
| Abstract | Multi-parameter magnetic resonance imaging (mp-MRI) is increasingly popular for prostate cancer (PCa) detection and diagnosis. However, interpreting mp-MRI data which typically contains multiple unregistered 3D sequences, e.g. apparent diffusion coefficient (ADC) and T2-weighted (T2w) images, is time-consuming and demands special expertise, limiting its usage for large-scale PCa screening. Therefore, solutions to computer-aided detection of PCa in mp-MRI images are highly desirable. Most recent advances in automated methods for PCa detection employ a handcrafted feature based two-stage classification flow, i.e. voxel-level classification followed by a region-level classification. This work presents an automated PCa detection system which can concurrently identify the presence of PCa in an image and localize lesions based on deep convolutional neural network (CNN) features and a single-stage SVM classifier. Specifically, the developed co-trained CNNs consist of two parallel convolutional networks for ADC and T2w images respectively. Each network is trained using images of a single modality in a weakly-supervised manner by providing a set of prostate images with image-level labels indicating only the presence of PCa without priors of lesions’ locations. Discriminative visual patterns of lesions can be learned effectively from clutters of prostate and surrounding tissues. A cancer response map with each pixel indicating the likelihood to be cancerous is explicitly generated at the last convolutional layer of the network for each modality. A new back-propagated error E is defined to enforce both optimized classification results and consistent cancer response maps for different modalities, which help capture highly representative PCa-relevant features during the CNN feature learning process. The CNN features of each modality are concatenated and fed into a SVM classifier. For images which are classified to contain cancers, non-maximum suppression and adaptive thresholding are applied to the corresponding cancer response maps for PCa foci localization. Evaluation based on 160 patient data with 12-core systematic TRUS-guided prostate biopsy as the reference standard demonstrates that our system achieves a sensitivity of 0.46, 0.92 and 0.97 at 0.1, 1 and 10 false positives per normal/benign patient which is significantly superior to two state-of-the-art CNN-based methods (Oquab et al., 2015; Zhou et al., 2015) and 6-core systematic prostate biopsies. |
| Tasks | |
| Published | 2017-08-24 |
| URL | https://www.ncbi.nlm.nih.gov/pubmed/28850876 |
| https://www.ncbi.nlm.nih.gov/pubmed/28850876 | |
| PWC | https://paperswithcode.com/paper/co-trained-convolutional-neural-networks-for |
| Repo | https://github.com/Andysis/co-trained-CADx |
| Framework | none |
Readers vs. Writers vs. Texts: Coping with Different Perspectives of Text Understanding in Emotion Annotation
| Title | Readers vs. Writers vs. Texts: Coping with Different Perspectives of Text Understanding in Emotion Annotation |
| Authors | Sven Buechel, Udo Hahn |
| Abstract | We here examine how different perspectives of understanding written discourse, like the reader{'}s, the writer{'}s or the text{'}s point of view, affect the quality of emotion annotations. We conducted a series of annotation experiments on two corpora, a popular movie review corpus and a genre- and domain-balanced corpus of standard English. We found statistical evidence that the writer{'}s perspective yields superior annotation quality overall. However, the quality one perspective yields compared to the other(s) seems to depend on the domain the utterance originates from. Our data further suggest that the popular movie review data set suffers from an atypical bimodal distribution which may decrease model performance when used as a training resource. |
| Tasks | Reading Comprehension |
| Published | 2017-04-01 |
| URL | https://www.aclweb.org/anthology/W17-0801/ |
| https://www.aclweb.org/anthology/W17-0801 | |
| PWC | https://paperswithcode.com/paper/readers-vs-writers-vs-texts-coping-with |
| Repo | https://github.com/JULIELab/EmoBank |
| Framework | none |
From Patches to Images: A Nonparametric Generative Model
| Title | From Patches to Images: A Nonparametric Generative Model |
| Authors | Geng Ji, Michael C. Hughes, Erik B. Sudderth |
| Abstract | We propose a hierarchical generative model that captures the self-similar structure of image regions as well as how this structure is shared across image collections. Our model is based on a novel, variational interpretation of the popular expected patch log-likelihood (EPLL) method as a model for randomly positioned grids of image patches. While previous EPLL methods modeled image patches with finite Gaussian mixtures, we use nonparametric Dirichlet process (DP) mixtures to create models whose complexity grows as additional images are observed. An extension based on the hierarchical DP then captures repetitive and self-similar structure via image-specific variations in cluster frequencies. We derive a structured variational inference algorithm that adaptively creates new patch clusters to more accurately model novel image textures. Our denoising performance on standard benchmarks is superior to EPLL and comparable to the state-of-the-art, and provides novel statistical justifications for common image processing heuristics. We also show accurate image inpainting results. |
| Tasks | Denoising, Image Inpainting |
| Published | 2017-08-01 |
| URL | https://icml.cc/Conferences/2017/Schedule?showEvent=753 |
| http://proceedings.mlr.press/v70/ji17a/ji17a.pdf | |
| PWC | https://paperswithcode.com/paper/from-patches-to-images-a-nonparametric |
| Repo | https://github.com/bnpy/hdp-grid-image-restoration |
| Framework | none |
``Liar, Liar Pants on Fire’': A New Benchmark Dataset for Fake News Detection
| Title | ``Liar, Liar Pants on Fire’': A New Benchmark Dataset for Fake News Detection | |
| Authors | William Yang Wang |
| Abstract | Automatic fake news detection is a challenging problem in deception detection, and it has tremendous real-world political and social impacts. However, statistical approaches to combating fake news has been dramatically limited by the lack of labeled benchmark datasets. In this paper, we present LIAR: a new, publicly available dataset for fake news detection. We collected a decade-long, 12.8K manually labeled short statements in various contexts from PolitiFact.com, which provides detailed analysis report and links to source documents for each case. This dataset can be used for fact-checking research as well. Notably, this new dataset is an order of magnitude larger than previously largest public fake news datasets of similar type. Empirically, we investigate automatic fake news detection based on surface-level linguistic patterns. We have designed a novel, hybrid convolutional neural network to integrate meta-data with text. We show that this hybrid approach can improve a text-only deep learning model. |
| Tasks | Deception Detection, Fake News Detection, Sentiment Analysis |
| Published | 2017-07-01 |
| URL | https://www.aclweb.org/anthology/P17-2067/ |
| https://www.aclweb.org/anthology/P17-2067 | |
| PWC | https://paperswithcode.com/paper/liar-liar-pants-on-fire-a-new-benchmark-1 |
| Repo | https://github.com/ekagra-ranjan/fake-news-detection-LIAR-pytorch |
| Framework | pytorch |
ESTEEM: A Novel Framework for Qualitatively Evaluating and Visualizing Spatiotemporal Embeddings in Social Media
| Title | ESTEEM: A Novel Framework for Qualitatively Evaluating and Visualizing Spatiotemporal Embeddings in Social Media |
| Authors | Dustin Arendt, Svitlana Volkova |
| Abstract | |
| Tasks | Named Entity Recognition, Part-Of-Speech Tagging, Topic Models, Word Embeddings |
| Published | 2017-07-01 |
| URL | https://www.aclweb.org/anthology/P17-4005/ |
| https://www.aclweb.org/anthology/P17-4005 | |
| PWC | https://paperswithcode.com/paper/esteem-a-novel-framework-for-qualitatively |
| Repo | https://github.com/pnnl/esteem |
| Framework | none |
ProtoNN: Compressed and Accurate kNN for Resource-scarce Devices
| Title | ProtoNN: Compressed and Accurate kNN for Resource-scarce Devices |
| Authors | Chirag Gupta, Arun Sai Suggala, Ankit Goyal, Harsha Vardhan Simhadri, Bhargavi Paranjape, Ashish Kumar, Saurabh Goyal, Raghavendra Udupa, Manik Varma, Prateek Jain |
| Abstract | Several real-world applications require real-time prediction on resource-scarce devices such as an Internet of Things (IoT) sensor. Such applications demand prediction models with small storage and computational complexity that do not compromise significantly on accuracy. In this work, we propose ProtoNN, a novel algorithm that addresses the problem of real-time and accurate prediction on resource-scarce devices. ProtoNN is inspired by k-Nearest Neighbor (KNN) but has several orders lower storage and prediction complexity. ProtoNN models can be deployed even on devices with puny storage and computational power (e.g. an Arduino UNO with 2kB RAM) to get excellent prediction accuracy. ProtoNN derives its strength from three key ideas: a) learning a small number of prototypes to represent the entire training set, b) sparse low dimensional projection of data, c) joint discriminative learning of the projection and prototypes with explicit model size constraint. We conduct systematic empirical evaluation of ProtoNN on a variety of supervised learning tasks (binary, multi-class, multi-label classification) and show that it gives nearly state-of-the-art prediction accuracy on resource-scarce devices while consuming several orders lower storage, and using minimal working memory. |
| Tasks | Multi-Label Classification |
| Published | 2017-08-01 |
| URL | https://icml.cc/Conferences/2017/Schedule?showEvent=683 |
| http://proceedings.mlr.press/v70/gupta17a/gupta17a.pdf | |
| PWC | https://paperswithcode.com/paper/protonn-compressed-and-accurate-knn-for |
| Repo | https://github.com/Microsoft/EdgeML |
| Framework | tf |
Centroid-based Text Summarization through Compositionality of Word Embeddings
| Title | Centroid-based Text Summarization through Compositionality of Word Embeddings |
| Authors | Gaetano Rossiello, Pierpaolo Basile, Giovanni Semeraro |
| Abstract | The textual similarity is a crucial aspect for many extractive text summarization methods. A bag-of-words representation does not allow to grasp the semantic relationships between concepts when comparing strongly related sentences with no words in common. To overcome this issue, in this paper we propose a centroid-based method for text summarization that exploits the compositional capabilities of word embeddings. The evaluations on multi-document and multilingual datasets prove the effectiveness of the continuous vector representation of words compared to the bag-of-words model. Despite its simplicity, our method achieves good performance even in comparison to more complex deep learning models. Our method is unsupervised and it can be adopted in other summarization tasks. |
| Tasks | Document Summarization, Multi-Document Summarization, Text Summarization, Word Embeddings |
| Published | 2017-04-01 |
| URL | https://www.aclweb.org/anthology/W17-1003/ |
| https://www.aclweb.org/anthology/W17-1003 | |
| PWC | https://paperswithcode.com/paper/centroid-based-text-summarization-through |
| Repo | https://github.com/gaetangate/text-summarizer |
| Framework | none |
CANE: Context-Aware Network Embedding for Relation Modeling
| Title | CANE: Context-Aware Network Embedding for Relation Modeling |
| Authors | Cunchao Tu, Han Liu, Zhiyuan Liu, Maosong Sun |
| Abstract | Network embedding (NE) is playing a critical role in network analysis, due to its ability to represent vertices with efficient low-dimensional embedding vectors. However, existing NE models aim to learn a fixed context-free embedding for each vertex and neglect the diverse roles when interacting with other vertices. In this paper, we assume that one vertex usually shows different aspects when interacting with different neighbor vertices, and should own different embeddings respectively. Therefore, we present Context-Aware Network Embedding (CANE), a novel NE model to address this issue. CANE learns context-aware embeddings for vertices with mutual attention mechanism and is expected to model the semantic relationships between vertices more precisely. In experiments, we compare our model with existing NE models on three real-world datasets. Experimental results show that CANE achieves significant improvement than state-of-the-art methods on link prediction and comparable performance on vertex classification. The source code and datasets can be obtained from \url{https://github.com/thunlp/CANE}. |
| Tasks | Community Detection, Link Prediction, Machine Translation, Network Embedding, Representation Learning |
| Published | 2017-07-01 |
| URL | https://www.aclweb.org/anthology/P17-1158/ |
| https://www.aclweb.org/anthology/P17-1158 | |
| PWC | https://paperswithcode.com/paper/cane-context-aware-network-embedding-for |
| Repo | https://github.com/thunlp/CANE |
| Framework | tf |
Region growing using superpixels with learned shape prior.
| Title | Region growing using superpixels with learned shape prior. |
| Authors | Jiří Borovec, Jan Kybic, Akihiro Sugimoto |
| Abstract | Region growing is a classical image segmentation method based on hierarchical region aggregation using local similarity rules. Our proposed method differs from classical region growing in three important aspects. First, it works on the level of superpixels instead of pixels, which leads to a substantial speed-up. Second, our method uses learned statistical shape properties that encourage plausible shapes. In particular, we use ray features to describe the object boundary. Third, our method can segment multiple objects and ensure that the segmentations do not overlap. The problem is represented as an energy minimization and is solved either greedily or iteratively using graph cuts. We demonstrate the performance of the proposed method and compare it with alternative approaches on the task of segmenting individual eggs in microscopy images of Drosophila ovaries. |
| Tasks | Semantic Segmentation |
| Published | 2017-11-06 |
| URL | https://doi.org/10.1117/1.JEI.26.6.061611 |
| https://doi.org/10.1117/1.JEI.26.6.061611 | |
| PWC | https://paperswithcode.com/paper/region-growing-using-superpixels-with-learned |
| Repo | https://github.com/Borda/pyImSegm |
| Framework | none |
