Predicting how events induce emotions in the characters of a story is typically seen as a standard multi-label classification task, which usually treats labels as anonymous classes to predict. They ignore information that may be conveyed by the emotion labels themselves. We propose that the semantics of emotion labels can guide a model's attention when representing the input story. Further, we observe that the emotions evoked by an event are often related: an event that evokes joy is unlikely to also evoke sadness. In this work, we explicitly model label classes via label embeddings, and add mechanisms that track label-label correlations both during training and inference. We also introduce a new semi-supervision strategy that regularizes for the correlations on unlabeled data. Our empirical evaluations show that modeling label semantics yields consistent benefits, and we advance the state-of-the-art on an emotion inference task.
Adversarial attacks on machine learning-based classifiers, along with defense mechanisms, have been widely studied in the context of single-label classification problems. In this paper, we shift the attention to multi-label classification, where the availability of domain knowledge on the relationships among the considered classes may offer a natural way to spot incoherent predictions, i.e., predictions associated to adversarial examples lying outside of the training data distribution. We explore this intuition in a framework in which first-order logic knowledge is converted into constraints and injected into a semi-supervised learning problem. Within this setting, the constrained classifier learns to fulfill the domain knowledge over the marginal distribution, and can naturally reject samples with incoherent predictions. Even though our method does not exploit any knowledge of attacks during training, our experimental analysis surprisingly unveils that domain-knowledge constraints can help detect adversarial examples effectively, especially if such constraints are not known to the attacker.
The outbreak of COVID-19 has shocked the entire world with its fairly rapid spread and has challenged different sectors. One of the most effective ways to limit its spread is the early and accurate diagnosing infected patients. Medical imaging, such as X-ray and Computed Tomography (CT), combined with the potential of Artificial Intelligence (AI), plays an essential role in supporting medical personnel in the diagnosis process. Thus, in this article five different deep learning models (ResNet18, ResNet34, InceptionV3, InceptionResNetV2 and DenseNet161) and their ensemble, using majority voting have been used to classify COVID-19, pneumoni{\ae} and healthy subjects using chest X-ray images. Multilabel classification was performed to predict multiple pathologies for each patient, if present. Firstly, the interpretability of each of the networks was thoroughly studied using local interpretability methods - occlusion, saliency, input X gradient, guided backpropagation, integrated gradients, and DeepLIFT, and using a global technique - neuron activation profiles. The mean Micro-F1 score of the models for COVID-19 classifications ranges from 0.66 to 0.875, and is 0.89 for the ensemble of the network models. The qualitative results showed that the ResNets were the most interpretable models. This research demonstrates the importance of using interpretability methods to compare different models before making a decision regarding the best performing model.
The skip-gram model for learning word embeddings (Mikolov et al. 2013) has been widely popular, and DeepWalk (Perozzi et al. 2014), among other methods, has extended the model to learning node representations from networks. Recent work of Qiu et al. (2018) provides a closed-form expression for the DeepWalk objective, obviating the need for sampling for small datasets and improving accuracy. In these methods, the "window size" T within which words or nodes are considered to co-occur is a key hyperparameter. We study the objective in the limit as T goes to infinity, which allows us to simplify the expression of Qiu et al. We prove that this limiting objective corresponds to factoring a simple transformation of the pseudoinverse of the graph Laplacian, linking DeepWalk to extensive prior work in spectral graph embeddings. Further, we show that by a applying a simple nonlinear entrywise transformation to this pseudoinverse, we recover a good approximation of the finite-T objective and embeddings that are competitive with those from DeepWalk and other skip-gram methods in multi-label classification. Surprisingly, we find that even simple binary thresholding of the Laplacian pseudoinverse is often competitive, suggesting that the core advancement of recent methods is a nonlinearity on top of the classical spectral embedding approach.
Movie genre classification is a challenging task that has increasingly attracted the attention of researchers. In this paper, we addressed the multi-label classification of the movie genres in a multimodal way. For this purpose, we created a dataset composed of trailer video clips, subtitles, synopses, and movie posters taken from 152,622 movie titles from The Movie Database. The dataset was carefully curated and organized, and it was also made available as a contribution of this work. Each movie of the dataset was labeled according to a set of eighteen genre labels. We extracted features from these data using different kinds of descriptors, namely Mel Frequency Cepstral Coefficients, Statistical Spectrum Descriptor , Local Binary Pattern with spectrograms, Long-Short Term Memory, and Convolutional Neural Networks. The descriptors were evaluated using different classifiers, such as BinaryRelevance and ML-kNN. We have also investigated the performance of the combination of different classifiers/features using a late fusion strategy, which obtained encouraging results. Based on the F-Score metric, our best result, 0.628, was obtained by the fusion of a classifier created using LSTM on the synopses, and a classifier created using CNN on movie trailer frames. When considering the AUC-PR metric, the best result, 0.673, was also achieved by combining those representations, but in addition, a classifier based on LSTM created from the subtitles was used. These results corroborate the existence of complementarity among classifiers based on different sources of information in this field of application. As far as we know, this is the most comprehensive study developed in terms of the diversity of multimedia sources of information to perform movie genre classification.
Multiple categories of objects are present in most images. Treating this as a multi-class classification is not justified. We treat this as a multi-label classification problem. In this paper, we further aim to minimize the supervision required for providing supervision in multi-label classification. Specifically, we investigate an effective class of approaches that associate a weak localization with each category either in terms of the bounding box or segmentation mask. Doing so improves the accuracy of multi-label categorization. The approach we adopt is one of active learning, i.e., incrementally selecting a set of samples that need supervision based on the current model, obtaining supervision for these samples, retraining the model with the additional set of supervised samples and proceeding again to select the next set of samples. A crucial concern is the choice of the set of samples. In doing so, we provide a novel insight, and no specific measure succeeds in obtaining a consistently improved selection criterion. We, therefore, provide a selection criterion that consistently improves the overall baseline criterion by choosing the top k set of samples for a varied set of criteria. Using this criterion, we are able to show that we can retain more than 98% of the fully supervised performance with just 20% of samples (and more than 96% using 10%) of the dataset on PASCAL VOC 2007 and 2012. Also, our proposed approach consistently outperforms all other baseline metrics for all benchmark datasets and model combinations.
In this paper, we show how selecting and combining encodings of natural and mathematical language affect classification and clustering of documents with mathematical content. We demonstrate this by using sets of documents, sections, and abstracts from the arXiv preprint server that are labeled by their subject class (mathematics, computer science, physics, etc.) to compare different encodings of text and formulae and evaluate the performance and runtimes of selected classification and clustering algorithms. Our encodings achieve classification accuracies up to $82.8\%$ and cluster purities up to $69.4\%$ (number of clusters equals number of classes), and $99.9\%$ (unspecified number of clusters) respectively. We observe a relatively low correlation between text and math similarity, which indicates the independence of text and formulae and motivates treating them as separate features of a document. The classification and clustering can be employed, e.g., for document search and recommendation. Furthermore, we show that the computer outperforms a human expert when classifying documents. Finally, we evaluate and discuss multi-label classification and formula semantification.
Automated Machine Learning (AutoML) has emerged to deal with the selection and configuration of algorithms for a given learning task. With the progression of AutoML, several effective methods were introduced, especially for traditional classification and regression problems. Apart from the AutoML success, several issues remain open. One issue, in particular, is the lack of ability of AutoML methods to deal with different types of data. Based on this scenario, this paper approaches AutoML for multi-label classification (MLC) problems. In MLC, each example can be simultaneously associated to several class labels, unlike the standard classification task, where an example is associated to just one class label. In this work, we provide a general comparison of five automated multi-label classification methods -- two evolutionary methods, one Bayesian optimization method, one random search and one greedy search -- on 14 datasets and three designed search spaces. Overall, we observe that the most prominent method is the one based on a canonical grammar-based genetic programming (GGP) search method, namely Auto-MEKA$_{GGP}$. Auto-MEKA$_{GGP}$ presented the best average results in our comparison and was statistically better than all the other methods in different search spaces and evaluated measures, except when compared to the greedy search method.
We present an elegant and effective approach for addressing limitations in existing multi-label classification models by incorporating interaction matching, a concept shown to be useful for ad-hoc search result ranking. By performing soft n-gram interaction matching, we match labels with natural language descriptions (which are common to have in most multi-labeling tasks). Our approach can be used to enhance existing multi-label classification approaches, which are biased toward frequently-occurring labels. We evaluate our approach on two challenging tasks: automatic medical coding of clinical notes and automatic labeling of entities from software tutorial text. Our results show that our method can yield up to an 11% relative improvement in macro performance, with most of the gains stemming labels that appear infrequently in the training set (i.e., the long tail of labels).
Videos convey rich information. Dynamic spatio-temporal relationships between people/objects, and diverse multimodal events are present in a video clip. Hence, it is important to develop automated models that can accurately extract such information from videos. Answering questions on videos is one of the tasks which can evaluate such AI abilities. In this paper, we propose a video question answering model which effectively integrates multi-modal input sources and finds the temporally relevant information to answer questions. Specifically, we first employ dense image captions to help identify objects and their detailed salient regions and actions, and hence give the model useful extra information (in explicit textual format to allow easier matching) for answering questions. Moreover, our model is also comprised of dual-level attention (word/object and frame level), multi-head self/cross-integration for different sources (video and dense captions), and gates which pass more relevant information to the classifier. Finally, we also cast the frame selection problem as a multi-label classification task and introduce two loss functions, In-andOut Frame Score Margin (IOFSM) and Balanced Binary Cross-Entropy (BBCE), to better supervise the model with human importance annotations. We evaluate our model on the challenging TVQA dataset, where each of our model components provides significant gains, and our overall model outperforms the state-of-the-art by a large margin (74.09% versus 70.52%). We also present several word, object, and frame level visualization studies. Our code is publicly available at: https://github.com/hyounghk/VideoQADenseCapFrameGate-ACL2020
Recently, sentiment analysis has seen remarkable advance with the help of pre-training approaches. However, sentiment knowledge, such as sentiment words and aspect-sentiment pairs, is ignored in the process of pre-training, despite the fact that they are widely used in traditional sentiment analysis approaches. In this paper, we introduce Sentiment Knowledge Enhanced Pre-training (SKEP) in order to learn a unified sentiment representation for multiple sentiment analysis tasks. With the help of automatically-mined knowledge, SKEP conducts sentiment masking and constructs three sentiment knowledge prediction objectives, so as to embed sentiment information at the word, polarity and aspect level into pre-trained sentiment representation. In particular, the prediction of aspect-sentiment pairs is converted into multi-label classification, aiming to capture the dependency between words in a pair. Experiments on three kinds of sentiment tasks show that SKEP significantly outperforms strong pre-training baseline, and achieves new state-of-the-art results on most of the test datasets. We release our code at https://github.com/baidu/Senta.
The identification of analytic tasks from free text is critical for visualization-oriented natural language interfaces (V-NLIs) to suggest effective visualizations. However, it is challenging due to the ambiguity and complexity nature of human language. To address this challenge, we present a new dataset, called Quda, that aims to help V-NLIs recognize analytic tasks from free-form natural language by training and evaluating cutting-edge multi-label classification models. Our dataset contains $14,035$ diverse user queries, and each is annotated with one or multiple analytic tasks. We achieve this goal by first gathering seed queries with data analysts and then employing extensive crowd force for paraphrase generation and validation. We demonstrate the usefulness of Quda through three applications. This work is the first attempt to construct a large-scale corpus for recognizing analytic tasks. With the release of Quda, we hope it will boost the research and development of V-NLIs in data analysis and visualization.
Following each patient visit, physicians draft long semi-structured clinical summaries called SOAP notes. While invaluable to clinicians and researchers, creating digital SOAP notes is burdensome, contributing to physician burnout. In this paper, we introduce the first complete pipelines to leverage deep summarization models to generate these notes based on transcripts of conversations between physicians and patients. After exploring a spectrum of methods across the extractive-abstractive spectrum, we propose Cluster2Sent, an algorithm that (i) extracts important utterances relevant to each summary section; (ii) clusters together related utterances; and then (iii) generates one summary sentence per cluster. Cluster2Sent outperforms its purely abstractive counterpart by 8 ROUGE-1 points, and produces significantly more factual and coherent sentences as assessed by expert human evaluators. For reproducibility, we demonstrate similar benefits on the publicly available AMI dataset. Our results speak to the benefits of structuring summaries into sections and annotating supporting evidence when constructing summarization corpora.
This work investigates the use of class-level difficulty factors in multi-label classification problems for the first time. Four class-level difficulty factors are proposed: frequency, visual variation, semantic abstraction, and class co-occurrence. Once computed for a given multi-label classification dataset, these difficulty factors are shown to have several potential applications including the prediction of class-level performance across datasets and the improvement of predictive performance through difficulty weighted optimisation. Significant improvements to mAP and AUC performance are observed for two challenging multi-label datasets (WWW Crowd and Visual Genome) with the inclusion of difficulty weighted optimisation. The proposed technique does not require any additional computational complexity during training or inference and can be extended over time with inclusion of other class-level difficulty factors.
The use of linguistic typological resources in natural language processing has been steadily gaining more popularity. It has been observed that the use of typological information, often combined with distributed language representations, leads to significantly more powerful models. While linguistic typology representations from various resources have mostly been used for conditioning the models, there has been relatively little attention on predicting features from these resources from the input data. In this paper we investigate whether the various linguistic features from World Atlas of Language Structures (WALS) can be reliably inferred from multi-lingual text. Such a predictor can be used to infer structural features for a language never observed in training data. We frame this task as a multi-label classification involving predicting the set of non-mutually exclusive and extremely sparse multi-valued labels (WALS features). We construct a recurrent neural network predictor based on byte embeddings and convolutional layers and test its performance on 556 languages, providing analysis for various linguistic types, macro-areas, language families and individual features. We show that some features from various linguistic types can be predicted reliably.
Pedestrian attribute recognition is an important multi-label classification problem. Although the convolutional neural networks are prominent in learning discriminative features from images, the data imbalance in multi-label setting for fine-grained tasks remains an open problem. In this paper, we propose a new re-sampling algorithm called: data augmentation imbalance (DAI) to explicitly enhance the ability to discriminate the fewer attributes via increasing the proportion of labels accounting for a small part. Fundamentally, by applying over-sampling and under-sampling on the multi-label dataset at the same time, the thought of robbing the rich attributes and helping the poor makes a significant contribution to DAI. Extensive empirical evidence shows that our DAI algorithm achieves state-of-the-art results, based on pedestrian attribute datasets, i.e. standard PA-100K and PETA datasets.
Open Knowledge Graphs (such as DBpedia, Wikidata, YAGO) have been recognized as the backbone of diverse applications in the field of data mining and information retrieval. Hence, the completeness and correctness of the Knowledge Graphs (KGs) are vital. Most of these KGs are mostly created either via an automated information extraction from Wikipedia snapshots or information accumulation provided by the users or using heuristics. However, it has been observed that the type information of these KGs is often noisy, incomplete, and incorrect. To deal with this problem a multi-label classification approach is proposed in this work for entity typing using KG embeddings. We compare our approach with the current state-of-the-art type prediction method and report on experiments with the KGs.
Multi-label networks with branches are proved to perform well in both accuracy and speed, but lacks flexibility in providing dynamic extension onto new labels due to the low efficiency of re-work on annotating and training. For multi-label classification task, to cover new labels we need to annotate not only newly collected images, but also the previous whole dataset to check presence of these new labels. Also training on whole re-annotated dataset costs much time. In order to recognize new labels more effectively and accurately, we propose GraftNet, which is a customizable tree-like network with its trunk pretrained with a dynamic graph for generic feature extraction, and branches separately trained on sub-datasets with single label to improve accuracy. GraftNet could reduce cost, increase flexibility, and incrementally handle new labels. Experimental results show that it has good performance on our human attributes recognition task, which is fine-grained multi-label classification.
In recent years, there has been an increasing interest in the application of Artificial Intelligence - and especially Machine Learning - to the field of Sustainable Development (SD). However, until now, NLP has not been applied in this context. In this research paper, we show the high potential of NLP applications to enhance the sustainability of projects. In particular, we focus on the case of community profiling in developing countries, where, in contrast to the developed world, a notable data gap exists. In this context, NLP could help to address the cost and time barrier of structuring qualitative data that prohibits its widespread use and associated benefits. We propose the new task of Automatic UPV classification, which is an extreme multi-class multi-label classification problem. We release Stories2Insights, an expert-annotated dataset, provide a detailed corpus analysis, and implement a number of strong neural baselines to address the task. Experimental results show that the problem is challenging, and leave plenty of room for future research at the intersection of NLP and SD.
Several learning algorithms have been proposed for offline multi-label classification. However, applications in areas such as traffic monitoring, social networks, and sensors produce data continuously, the so called data streams, posing challenges to batch multi-label learning. With the lack of stationarity in the distribution of data streams, new algorithms are needed to online adapt to such changes (concept drift). Also, in realistic applications, changes occur in scenarios of infinitely delayed labels, where the true classes of the arrival instances are never available. We propose an online unsupervised incremental method based on self-organizing maps for multi-label stream classification with infinitely delayed labels. In the classification phase, we use a k-nearest neighbors strategy to compute the winning neurons in the maps, adapting to concept drift by online adjusting neuron weight vectors and dataset label cardinality. We predict labels for each instance using the Bayes rule and the outputs of each neuron, adapting the probabilities and conditional probabilities of the classes in the stream. Experiments using synthetic and real datasets show that our method is highly competitive with several ones from the literature, in both stationary and concept drift scenarios.