The performance of text classification methods has improved greatly over the last decade for text instances of less than 512 tokens. This limit has been adopted by most state-of-the-research transformer models due to the high computational cost of analyzing longer text instances. To mitigate this problem and to improve classification for longer texts, researchers have sought to resolve the underlying causes of the computational cost and have proposed optimizations for the attention mechanism, which is the key element of every transformer model. In our study, we are not pursuing the ultimate goal of long text classification, i.e., the ability to analyze entire text instances at one time while preserving high performance at a reasonable computational cost. Instead, we propose a text truncation method called Text Guide, in which the original text length is reduced to a predefined limit in a manner that improves performance over naive and semi-naive approaches while preserving low computational costs. Text Guide benefits from the concept of feature importance, a notion from the explainable artificial intelligence domain. We demonstrate that Text Guide can be used to improve the performance of recent language models specifically designed for long text classification, such as Longformer. Moreover, we discovered that parameter optimization is the key to Text Guide performance and must be conducted before the method is deployed. Future experiments may reveal additional benefits provided by this new method.
In this paper we analyze the classification performance of neural network structures without parametric inference. Making use of neural architecture search, we empirically demonstrate that it is possible to find random weight architectures, a deep prior, that enables a linear classification to perform on par with fully trained deep counterparts. Through ablation experiments, we exclude the possibility of winning a weight initialization lottery and confirm that suitable deep priors do not require additional inference. In an extension to continual learning, we investigate the possibility of catastrophic interference free incremental learning. Under the assumption of classes originating from the same data distribution, a deep prior found on only a subset of classes is shown to allow discrimination of further classes through training of a simple linear classifier.
Graph Convolutional Networks (GCNs) have been extensively used to classify vertices in graphs and have been shown to outperform other vertex classification methods. GCNs have been extended to graph classification tasks (GCT). In GCT, graphs with different numbers of edges and vertices belong to different classes, and one attempts to predict the graph class. GCN based GCT have mostly used pooling and attention-based models. The accuracy of existing GCT methods is still limited. We here propose a novel solution combining GCN, methods from knowledge graphs, and a new self-regularized activation function to significantly improve the accuracy of the GCN based GCT. We present quadratic GCN (QGCN) - A GCN formalism with a quadratic layer. Such a layer produces an output with fixed dimensions, independent of the graph vertex number. We applied this method to a wide range of graph classification problems, and show that when using a self regularized activation function, QGCN outperforms the state of the art methods for all graph classification tasks tested with or without external input on each graph. The code for QGCN is available at: https://github.com/Unknown-Data/QGCN .
Purpose: we evaluated the generalization capability of deep neural networks (DNNs), trained to classify chest X-rays as Covid-19, normal or pneumonia, using a relatively small and mixed dataset. Methods: we proposed a DNN to perform lung segmentation and classification, stacking a segmentation module (U-Net), an original intermediate module and a classification module (DenseNet201). To evaluate generalization, we tested the DNN with an external dataset (from distinct localities) and used Bayesian inference to estimate probability distributions of performance metrics. Results: our DNN achieved 0.917 AUC on the external test dataset, and a DenseNet without segmentation, 0.906. Bayesian inference indicated mean accuracy of 76.1% and [0.695, 0.826] 95% HDI (highest density interval, which concentrates 95% of the metric's probability mass) with segmentation and, without segmentation, 71.7% and [0.646, 0.786]. Conclusion: employing a novel DNN evaluation technique, which uses LRP and Brixia scores, we discovered that areas where radiologists found strong Covid-19 symptoms are the most important for the stacked DNN classification. External validation showed smaller accuracies than internal, indicating difficulty in generalization, which is positively affected by segmentation. Finally, the performance in the external dataset and the analysis with LRP suggest that DNNs can be trained in small and mixed datasets and still successfully detect Covid-19.
Contrastive learning has delivered impressive results for various tasks in the self-supervised regime. However, existing approaches optimize for learning representations specific to downstream scenarios, i.e., \textit{global} representations suitable for tasks such as classification or \textit{local} representations for tasks such as detection and localization. While they produce satisfactory results in the intended downstream scenarios, they often fail to generalize to tasks that they were not originally designed for. In this work, we propose to learn video representations that generalize to both the tasks which require global semantic information (e.g., classification) and the tasks that require local fine-grained spatio-temporal information (e.g., localization). We achieve this by optimizing two contrastive objectives that together encourage our model to learn global-local visual information given audio signals. We show that the two objectives mutually improve the generalizability of the learned global-local representations, significantly outperforming their disjointly learned counterparts. We demonstrate our approach on various tasks including action/sound classification, lip reading, deepfake detection, event and sound localization (https://github.com/yunyikristy/global\_local).
Neural topic models can augment or replace bag-of-words inputs with the learned representations of deep pre-trained transformer-based word prediction models. One added benefit when using representations from multilingual models is that they facilitate zero-shot polylingual topic modeling. However, while it has been widely observed that pre-trained embeddings should be fine-tuned to a given task, it is not immediately clear what supervision should look like for an unsupervised task such as topic modeling. Thus, we propose several methods for fine-tuning encoders to improve both monolingual and zero-shot polylingual neural topic modeling. We consider fine-tuning on auxiliary tasks, constructing a new topic classification task, integrating the topic classification objective directly into topic model training, and continued pre-training. We find that fine-tuning encoder representations on topic classification and integrating the topic classification task directly into topic modeling improves topic quality, and that fine-tuning encoder representations on any task is the most important factor for facilitating cross-lingual transfer.
The UNESCO World Heritage List (WHL) includes the exceptionally valuable cultural and natural heritage to be preserved for mankind. Evaluating and justifying the Outstanding Universal Value (OUV) is essential for each site inscribed in the WHL, and yet a complex task, even for experts, since the selection criteria of OUV are not mutually exclusive. Furthermore, manual annotation of heritage values and attributes from multi-source textual data, which is currently dominant in heritage studies, is knowledge-demanding and time-consuming, impeding systematic analysis of such authoritative documents in terms of their implications on heritage management. This study applies state-of-the-art NLP models to build a classifier on a new dataset containing Statements of OUV, seeking an explainable and scalable automation tool to facilitate the nomination, evaluation, research, and monitoring processes of World Heritage sites. Label smoothing is innovatively adapted to improve the model performance by adding prior inter-class relationship knowledge to generate soft labels. The study shows that the best models fine-tuned from BERT and ULMFiT can reach 94.3% top-3 accuracy. A human study with expert evaluation on the model prediction shows that the models are sufficiently generalizable. The study is promising to be further developed and applied in heritage research and practice.
The current state-of-the-art model HiAGM for hierarchical text classification has two limitations. First, it correlates each text sample with all labels in the dataset which contains irrelevant information. Second, it does not consider any statistical constraint on the label representations learned by the structure encoder, while constraints for representation learning are proved to be helpful in previous work. In this paper, we propose HTCInfoMax to address these issues by introducing information maximization which includes two modules: text-label mutual information maximization and label prior matching. The first module can model the interaction between each text sample and its ground truth labels explicitly which filters out irrelevant information. The second one encourages the structure encoder to learn better representations with desired characteristics for all labels which can better handle label imbalance in hierarchical text classification. Experimental results on two benchmark datasets demonstrate the effectiveness of the proposed HTCInfoMax.
Contrastive Learning has emerged as a powerful representation learning method and facilitates various downstream tasks especially when supervised data is limited. How to construct efficient contrastive samples through data augmentation is key to its success. Unlike vision tasks, the data augmentation method for contrastive learning has not been investigated sufficiently in language tasks. In this paper, we propose a novel approach to construct contrastive samples for language tasks using text summarization. We use these samples for supervised contrastive learning to gain better text representations which greatly benefit text classification tasks with limited annotations. To further improve the method, we mix up samples from different classes and add an extra regularization, named Mixsum, in addition to the cross-entropy-loss. Experiments on real-world text classification datasets (Amazon-5, Yelp-5, AG News, and IMDb) demonstrate the effectiveness of the proposed contrastive learning framework with summarization-based data augmentation and Mixsum regularization.
Contemporary Artificial Intelligence technologies allow for the employment of Computer Vision to discern good crops from bad, providing a step in the pipeline of selecting healthy fruit from undesirable fruit, such as those which are mouldy or gangrenous. State-of-the-art works in the field report high accuracy results on small datasets (<1000 images), which are not representative of the population regarding real-world usage. The goals of this study are to further enable real-world usage by improving generalisation with data augmentation as well as to reduce overfitting and energy usage through model pruning. In this work, we suggest a machine learning pipeline that combines the ideas of fine-tuning, transfer learning, and generative model-based training data augmentation towards improving fruit quality image classification. A linear network topology search is performed to tune a VGG16 lemon quality classification model using a publicly-available dataset of 2690 images. We find that appending a 4096 neuron fully connected layer to the convolutional layers leads to an image classification accuracy of 83.77%. We then train a Conditional Generative Adversarial Network on the training data for 2000 epochs, and it learns to generate relatively realistic images. Grad-CAM analysis of the model trained on real photographs shows that the synthetic images can exhibit classifiable characteristics such as shape, mould, and gangrene. A higher image classification accuracy of 88.75% is then attained by augmenting the training with synthetic images, arguing that Conditional Generative Adversarial Networks have the ability to produce new data to alleviate issues of data scarcity. Finally, model pruning is performed via polynomial decay, where we find that the Conditional GAN-augmented classification network can retain 81.16% classification accuracy when compressed to 50% of its original size.
We propose a two-stage Convolutional Neural Network (CNN) based classification framework for detecting COVID-19 and Community-Acquired Pneumonia (CAP) using the chest Computed Tomography (CT) scan images. In the first stage, an infection - COVID-19 or CAP, is detected using a pre-trained DenseNet architecture. Then, in the second stage, a fine-grained three-way classification is done using EfficientNet architecture. The proposed COVID+CAP-CNN framework achieved a slice-level classification accuracy of over 94% at identifying COVID-19 and CAP. Further, the proposed framework has the potential to be an initial screening tool for differential diagnosis of COVID-19 and CAP, achieving a validation accuracy of over 89.3% at the finer three-way COVID-19, CAP, and healthy classification. Within the IEEE ICASSP 2021 Signal Processing Grand Challenge (SPGC) on COVID-19 Diagnosis, our proposed two-stage classification framework achieved an overall accuracy of 90% and sensitivity of .857, .9, and .942 at distinguishing COVID-19, CAP, and normal individuals respectively, to rank first in the evaluation. Code and model weights are available at https://github.com/shubhamchaudhary2015/ct_covid19_cap_cnn
Zero-shot learning, the task of learning to recognize new classes not seen during training, has received considerable attention in the case of 2D image classification. However, despite the increasing ubiquity of 3D sensors, the corresponding 3D point cloud classification problem has not been meaningfully explored and introduces new challenges. In this paper, we identify some of the challenges and apply 2D Zero-Shot Learning (ZSL) methods in the 3D domain to analyze the performance of existing models. Then, we propose a novel approach to address the issues specific to 3D ZSL. We first present an inductive ZSL process and then extend it to the transductive ZSL and Generalized ZSL (GZSL) settings for 3D point cloud classification. To this end, a novel loss function is developed that simultaneously aligns seen semantics with point cloud features and takes advantage of unlabeled test data to address some known issues (e.g., the problems of domain adaptation, hubness, and data bias). While designed for the particularities of 3D point cloud classification, the method is shown to also be applicable to the more common use-case of 2D image classification. An extensive set of experiments is carried out, establishing state-of-the-art for ZSL and GZSL on synthetic (ModelNet40, ModelNet10, McGill) and real (ScanObjectNN) 3D point cloud datasets.
Continual learning has become increasingly important as it enables NLP models to constantly learn and gain knowledge over time. Previous continual learning methods are mainly designed to preserve knowledge from previous tasks, without much emphasis on how to well generalize models to new tasks. In this work, we propose an information disentanglement based regularization method for continual learning on text classification. Our proposed method first disentangles text hidden spaces into representations that are generic to all tasks and representations specific to each individual task, and further regularizes these representations differently to better constrain the knowledge required to generalize. We also introduce two simple auxiliary tasks: next sentence prediction and task-id prediction, for learning better generic and specific representation spaces. Experiments conducted on large-scale benchmarks demonstrate the effectiveness of our method in continual text classification tasks with various sequences and lengths over state-of-the-art baselines. We have publicly released our code at https://github.com/GT-SALT/IDBR.
Lipschitz constrained networks have gathered considerable attention in the deep learning community, with usages ranging from Wasserstein distance estimation to the training of certifiably robust classifiers. However they remain commonly considered as less accurate, and their properties in learning are still not fully understood. In this paper we clarify the matter: when it comes to classification 1-Lipschitz neural networks enjoy several advantages over their unconstrained counterpart. First, we show that these networks are as accurate as classical ones, and can fit arbitrarily difficult boundaries. Then, relying on a robustness metric that reflects operational needs we characterize the most robust classifier: the WGAN discriminator. Next, we show that 1-Lipschitz neural networks generalize well under milder assumptions. Finally, we show that hyper-parameters of the loss are crucial for controlling the accuracy-robustness trade-off. We conclude that they exhibit appealing properties to pave the way toward provably accurate, and provably robust neural networks.
Class imbalance is an inherent problem in many machine learning classification tasks. This often leads to trained models that are unusable for any practical purpose. In this study we explore an unsupervised approach to address these imbalances by leveraging transfer learning from pre-trained image classification models to encoder-based Generative Adversarial Network (eGAN). To the best of our knowledge, this is the first work to tackle this problem using GAN without needing to augment with synthesized fake images. In the proposed approach we use the discriminator network to output a negative or positive score. We classify as minority, test samples with negative scores and as majority those with positive scores. Our approach eliminates epistemic uncertainty in model predictions, as the P(minority) + P(majority) need not sum up to 1. The impact of transfer learning and combinations of different pre-trained image classification models at the generator and discriminator is also explored. Best result of 0.69 F1-score was obtained on CIFAR-10 classification task with imbalance ratio of 1:2500. Our approach also provides a mechanism of thresholding the specificity or sensitivity of our machine learning system. Keywords: Class imbalance, Transfer Learning, GAN, nash equilibrium
Text classification is a significant branch of natural language processing, and has many applications including document classification and sentiment analysis. Unsurprisingly, those who do text classification are concerned with the run-time of their algorithms, many of which depend on the size of the corpus' vocabulary due to their bag-of-words representation. Although many studies have examined the effect of preprocessing techniques on vocabulary size and accuracy, none have examined how these methods affect a model's run-time. To fill this gap, we provide a comprehensive study that examines how preprocessing techniques affect the vocabulary size, model performance, and model run-time, evaluating ten techniques over four models and two datasets. We show that some individual methods can reduce run-time with no loss of accuracy, while some combinations of methods can trade 2-5% of the accuracy for up to a 65% reduction of run-time. Furthermore, some combinations of preprocessing techniques can even provide a 15% reduction in run-time while simultaneously improving model accuracy.
Recent research has shown the great potential of deep learning algorithms in the hyperspectral image (HSI) classification task. Nevertheless, training these models usually requires a large amount of labeled data. Since the collection of pixel-level annotations for HSI is laborious and time-consuming, developing algorithms that can yield good performance in the small sample size situation is of great significance. In this study, we propose a robust self-ensembling network (RSEN) to address this problem. The proposed RSEN consists of two subnetworks including a base network and an ensemble network. With the constraint of both the supervised loss from the labeled data and the unsupervised loss from the unlabeled data, the base network and the ensemble network can learn from each other, achieving the self-ensembling mechanism. To the best of our knowledge, the proposed method is the first attempt to introduce the self-ensembling technique into the HSI classification task, which provides a different view on how to utilize the unlabeled data in HSI to assist the network training. We further propose a novel consistency filter to increase the robustness of self-ensembling learning. Extensive experiments on three benchmark HSI datasets demonstrate that the proposed algorithm can yield competitive performance compared with the state-of-the-art methods. Code is available online (\url{https://github.com/YonghaoXu/RSEN}).
This paper presents DeepI2P: a novel approach for cross-modality registration between an image and a point cloud. Given an image (e.g. from a rgb-camera) and a general point cloud (e.g. from a 3D Lidar scanner) captured at different locations in the same scene, our method estimates the relative rigid transformation between the coordinate frames of the camera and Lidar. Learning common feature descriptors to establish correspondences for the registration is inherently challenging due to the lack of appearance and geometric correlations across the two modalities. We circumvent the difficulty by converting the registration problem into a classification and inverse camera projection optimization problem. A classification neural network is designed to label whether the projection of each point in the point cloud is within or beyond the camera frustum. These labeled points are subsequently passed into a novel inverse camera projection solver to estimate the relative pose. Extensive experimental results on Oxford Robotcar and KITTI datasets demonstrate the feasibility of our approach. Our source code is available at https://github.com/lijx10/DeepI2P
Cloth categorization is an important research problem that is used by e-commerce websites for displaying correct products to the end-users. Indian clothes have a large number of clothing categories both for men and women. The traditional Indian clothes like "Saree" and "Dhoti" are worn very differently from western clothes like t-shirts and jeans. Moreover, the style and patterns of ethnic clothes have a very different distribution from western outfits. Thus the models trained on standard cloth datasets fail miserably on ethnic outfits. To address these challenges, we introduce the first large-scale ethnic dataset of over 106k images with 15 different categories for fine-grained classification of Indian ethnic clothes. We gathered a diverse dataset from a large number of Indian e-commerce websites. We then evaluate several baselines for the cloth classification task on our dataset. In the end, we obtain 88.43% classification accuracy. We hope that our dataset would foster research in the development of several algorithms such as cloth classification, landmark detection, especially for ethnic clothes.
Knowledge distillation (KD) has been actively studied for image classification tasks in deep learning, aiming to improve the performance of a student based on the knowledge from a teacher. However, applying KD in image regression with a scalar response variable has been rarely studied, and there exists no KD method applicable to both classification and regression tasks yet. Moreover, existing KD methods often require a practitioner to carefully select or adjust the teacher and student architectures, making these methods less flexible in practice. To address the above problems in a unified way, we propose a comprehensive KD framework based on cGANs, termed cGAN-KD. Fundamentally different from existing KD methods, cGAN-KD distills and transfers knowledge from a teacher model to a student model via cGAN-generated samples. This novel mechanism makes cGAN-KD suitable for both classification and regression tasks, compatible with other KD methods, and insensitive to the teacher and student architectures. An error bound for a student model trained in the cGAN-KD framework is derived in this work, providing a theory for why cGAN-KD is effective as well as guiding the practical implementation of cGAN-KD. Extensive experiments on CIFAR-100 and ImageNet-100 show that we can combine state of the art KD methods with the cGAN-KD framework to yield a new state of the art. Moreover, experiments on Steering Angle and UTKFace demonstrate the effectiveness of cGAN-KD in image regression tasks, where existing KD methods are inapplicable.