Search

Invoking the manifold assumption in machine learning requires knowledge of the manifold's geometry and dimension, and theory dictates how many samples are required. However, in applications data are limited, sampling may not be uniform, and manifold properties are unknown and (possibly) non-pure; this implies that neighborhoods must adapt to the local structure. We introduce an algorithm for inferring adaptive neighborhoods for data given by a similarity kernel. Starting with a locally-conservative neighborhood (Gabriel) graph, we sparsify it iteratively according to a weighted counterpart. In each step, a linear program yields minimal neighborhoods globally and a volumetric statistic reveals neighbor outliers likely to violate manifold geometry. We apply our adaptive neighborhoods to non-linear dimensionality reduction, geodesic computation and dimension estimation. A comparison against standard algorithms using, e.g., k-nearest neighbors, demonstrates their usefulness. Code for our algorithm will be available at https://github.com/dyballa/IAN, 69 pages, 35 figures

Continual learning models for stationary data focus on learning and retaining concepts coming to them in a sequential manner. In the most generic class-incremental environment, we have to be ready to deal with classes coming one by one, without any higher-level grouping. This requirement invalidates many previously proposed methods and forces researchers to look for more flexible alternative approaches. In this work, we follow the idea of centroid-driven methods and propose end-to-end incorporation of the mixture of Gaussians model into the continual learning framework. By employing the gradient-based approach and designing losses capable of learning discriminative features while avoiding degenerate solutions, we successfully combine the mixture model with a deep feature extractor allowing for joint optimization and adjustments in the latent space. Additionally, we show that our model can effectively learn in memory-free scenarios with fixed extractors. In the conducted experiments, we empirically demonstrate the effectiveness of the proposed solutions and exhibit the competitiveness of our model when compared with state-of-the-art continual learning baselines evaluated in the context of image classification problems.

Du e to rapid population growth and the need to use artificial intelligence to make quick decisions, developing a machine learning-based disease detection model and abnormality identification system has greatly improved the level of medical diagnosis Since COVID-19 has become one of the most severe diseases in the world, developing an automatic COVID-19 detection framework helps medical doctors in the diagnostic process of disease and provides correct and fast results. In this paper, we propose a machine lear ning based framework for the detection of Covid 19. The proposed model employs a Tsukamoto Neuro Fuzzy Inference network to identify and distinguish Covid 19 disease from normal and pneumonia cases. While the traditional training methods tune the parameters of the neuro-fuzzy model by gradient-based algorithms and recursive least square method, we use an evolutionary-based optimization, the Cat swarm algorithm to update the parameters. In addition, six texture features extracted from chest X-ray images are give n as input to the model. Finally, the proposed model is conducted on the chest X-ray dataset to detect Covid 19. The simulation results indicate that the proposed model achieves an accuracy of 98.51%, sensitivity of 98.35%, specificity of 98.08%, and F1 score of 98.17%., 14 pages, 5 figures, 3 tables

The ACM Multimedia 2023 Computational Paralinguistics Challenge addresses two different problems for the first time in a research competition under well-defined conditions: In the Emotion Share Sub-Challenge, a regression on speech has to be made; and in the Requests Sub-Challenges, requests and complaints need to be detected. We describe the Sub-Challenges, baseline feature extraction, and classifiers based on the usual ComPaRE features, the auDeep toolkit, and deep feature extraction from pre-trained CNNs using the DeepSpectRum toolkit; in addition, wav2vec2 models are used., Comment: 5 pages, part of the ACM Multimedia 2023 Grand Challenge "The ACM Multimedia 2023 Computational Paralinguistics Challenge (ComParE 2023). arXiv admin note: text overlap with arXiv:2205.06799

The detection and localization of possible diseases in crops are usually automated by resorting to supervised deep learning approaches. In this work, we tackle these goals with unsupervised models, by applying three different types of autoencoders to a specific open-source dataset of healthy and unhealthy pepper and cherry leaf images. CAE, CVAE and VQ-VAE autoencoders are deployed to screen unlabeled images of such a dataset, and compared in terms of image reconstruction, anomaly removal, detection and localization. The vector-quantized variational architecture turns out to be the best performing one with respect to all these targets., 23 pages, 8 figures

In many scenarios, it is necessary to monitor a complex system via a time-series of observations and determine when anomalous exogenous events have occurred so that relevant actions can be taken. Determining whether current observations are abnormal is challenging. It requires learning an extrapolative probabilistic model of the dynamics from historical data, and using a limited number of current observations to make a classification. We leverage recent advances in long-term probabilistic forecasting, namely {\em Deep Probabilistic Koopman}, to build a general method for classifying anomalies in multi-dimensional time-series data. We also show how to utilize models with domain knowledge of the dynamics to reduce type I and type II error. We demonstrate our proposed method on the important real-world task of global atmospheric pollution monitoring, integrating it with NASA's Global Earth System Model. The system successfully detects localized anomalies in air quality due to events such as COVID-19 lockdowns and wildfires., 10 pages, 8 figures

The brains of all bilaterally symmetric animals on Earth are divided into left and right hemispheres. The anatomy and functionality of the hemispheres have a large degree of overlap, but there are asymmetries and they specialize to possess different attributes. Several studies have used computational models to mimic hemispheric asymmetries with a focus on reproducing human data on semantic and visual processing tasks. In this study, we aimed to understand how dual hemispheres could interact in a given task. We propose a bilateral artificial neural network that imitates a lateralization observed in nature: that the left hemisphere specializes in specificity and the right in generalities. We used two ResNet-9 convolutional neural networks with different training objectives and tested it on an image classification task. Our analysis found that the hemispheres represent complementary features that are exploited by a network head which implements a type of weighted attention. The bilateral architecture outperformed a range of baselines of similar representational capacity that don't exploit differential specialization. The results demonstrate the efficacy of bilateralism, contribute to an understanding of bilateralism in biological brains and the architecture serves as an inductive bias when designing new AI systems., 14 pages, 11 figures

Continual learning and few-shot learning are important frontiers in the quest to improve Machine Learning. There is a growing body of work in each frontier, but very little combining the two. Recently however, Antoniou et al. arXiv:2004.11967 introduced a Continual Few-shot Learning framework, CFSL, that combines both. In this study, we extended CFSL to make it more comparable to standard continual learning experiments, where usually a much larger number of classes are presented. We also introduced an `instance test' to classify very similar specific instances - a capability of animal cognition that is usually neglected in ML. We selected representative baseline models from the original CFSL work and compared to a model with Hippocampal-inspired replay, as the Hippocampus is considered to be vital to this type of learning in animals. As expected, learning more classes is more difficult than the original CFSL experiments, and interestingly, the way in which they are presented makes a difference to performance. Accuracy in the instance test is comparable to the classification tasks. The use of replay for consolidation improves performance substantially for both types of tasks, particularly the instance test., 11 pages, 5 figures

Subgroup discovery is a descriptive and exploratory data mining technique to identify subgroups in a population that exhibit interesting behavior with respect to a variable of interest. Subgroup discovery has numerous applications in knowledge discovery and hypothesis generation, yet it remains inapplicable for unstructured, high-dimensional data such as images. This is because subgroup discovery algorithms rely on defining descriptive rules based on (attribute, value) pairs, however, in unstructured data, an attribute is not well defined. Even in cases where the notion of attribute intuitively exists in the data, such as a pixel in an image, due to the high dimensionality of the data, these attributes are not informative enough to be used in a rule. In this paper, we introduce the subgroup-aware variational autoencoder, a novel variational autoencoder that learns a representation of unstructured data which leads to subgroups with higher quality. Our experimental results demonstrate the effectiveness of the method at learning subgroups with high quality while supporting the interpretability of the concepts.

This paper proposes a method to accelerate the training process of a general fuzzy min-max neural network. The purpose is to reduce the unsuitable hyperboxes selected as the potential candidates of the expansion step of existing hyperboxes to cover a new input pattern in the online learning algorithms or candidates of the hyperbox aggregation process in the agglomerative learning algorithms. Our proposed approach is based on the mathematical formulas to form a branch-and-bound solution aiming to remove the hyperboxes which are certain not to satisfy expansion or aggregation conditions, and in turn, decreasing the training time of learning algorithms. The efficiency of the proposed method is assessed over a number of widely used data sets. The experimental results indicated the significant decrease in training time of the proposed approach for both online and agglomerative learning algorithms. Notably, the training time of the online learning algorithms is reduced from 1.2 to 12 times when using the proposed method, while the agglomerative learning algorithms are accelerated from 7 to 37 times on average.

Modern applications such as voice recognition rely on the ability to compare signals to pre-recorded ones to classify them. However, this comparison typically needs to ignore differences due to signal noise, temporal offset, signal magnitude, and other external factors. The Dynamic Time Warping (DTW) algorithm quantifies this similarity by finding corresponding regions between the signals and non-linearly warping one signal by stretching and shrinking it. Unfortunately, searching through all "warps" of a signal to find the best corresponding regions is computationally expensive. The FastDTW algorithm improves performance, but sacrifices accuracy by only considering small signal warps. My goal is to improve the speed of DTW while maintaining high accuracy. My key insight is that in any particular application domain, signals exhibit specific types of variation. For example, the accelerometer signal measured for two different people would differ based on their stride length and weight. My system, called Machine Learning DTW (MLDTW), uses machine learning to learn the types of warps that are common in a particular domain. It then uses the learned model to improve DTW performance by limiting the search of potential warps appropriately. My results show that compared to FastDTW, MLDTW is at least as fast and reduces errors by 60% on average across four different data sets. These improvements will significantly impact a wide variety of applications (e.g. health monitoring) and enable more scalable processing of multivariate, higher frequency, and longer signal recordings., Presented at Regeneron International Science and Engineering Fair (ISEF) 2022

Continuous learning from streaming data is among the most challenging topics in the contemporary machine learning. In this domain, learning algorithms must not only be able to handle massive volumes of rapidly arriving data, but also adapt themselves to potential emerging changes. The phenomenon of the evolving nature of data streams is known as concept drift. While there is a plethora of methods designed for detecting its occurrence, all of them assume that the drift is connected with underlying changes in the source of data. However, one must consider the possibility of a malicious injection of false data that simulates a concept drift. This adversarial setting assumes a poisoning attack that may be conducted in order to damage the underlying classification system by forcing adaptation to false data. Existing drift detectors are not capable of differentiating between real and adversarial concept drift. In this paper, we propose a framework for robust concept drift detection in the presence of adversarial and poisoning attacks. We introduce the taxonomy for two types of adversarial concept drifts, as well as a robust trainable drift detector. It is based on the augmented Restricted Boltzmann Machine with improved gradient computation and energy function. We also introduce Relative Loss of Robustness - a novel measure for evaluating the performance of concept drift detectors under poisoning attacks. Extensive computational experiments, conducted on both fully and sparsely labeled data streams, prove the high robustness and efficacy of the proposed drift detection framework in adversarial scenarios., 42 pages, 13 figures, 8 tables

As automated machine learning (AutoML) systems continue to progress in both sophistication and performance, it becomes important to understand the `how' and `why' of human-computer interaction (HCI) within these frameworks, both current and expected. Such a discussion is necessary for optimal system design, leveraging advanced data-processing capabilities to support decision-making involving humans, but it is also key to identifying the opportunities and risks presented by ever-increasing levels of machine autonomy. Within this context, we focus on the following questions: (i) How does HCI currently look like for state-of-the-art AutoML algorithms, especially during the stages of development, deployment, and maintenance? (ii) Do the expectations of HCI within AutoML frameworks vary for different types of users and stakeholders? (iii) How can HCI be managed so that AutoML solutions acquire human trust and broad acceptance? (iv) As AutoML systems become more autonomous and capable of learning from complex open-ended environments, will the fundamental nature of HCI evolve? To consider these questions, we project existing literature in HCI into the space of AutoML; this connection has, to date, largely been unexplored. In so doing, we review topics including user-interface design, human-bias mitigation, and trust in artificial intelligence (AI). Additionally, to rigorously gauge the future of HCI, we contemplate how AutoML may manifest in effectively open-ended environments. This discussion necessarily reviews projected developmental pathways for AutoML, such as the incorporation of reasoning, although the focus remains on how and why HCI may occur in such a framework rather than on any implementational details. Ultimately, this review serves to identify key research directions aimed at better facilitating the roles and modes of human interactions with both current and future AutoML systems., Submitted to Foundations and Trends in Human-Computer Interaction

General fuzzy min-max (GFMM) neural network is a generalization of fuzzy neural networks formed by hyperbox fuzzy sets for classification and clustering problems. Two principle algorithms are deployed to train this type of neural network, i.e., incremental learning and agglomerative learning. This paper presents a comprehensive empirical study of performance influencing factors, advantages, and drawbacks of the general fuzzy min-max neural network on pattern classification problems. The subjects of this study include (1) the impact of maximum hyperbox size, (2) the influence of the similarity threshold and measures on the agglomerative learning algorithm, (3) the effect of data presentation order, (4) comparative performance evaluation of the GFMM with other types of fuzzy min-max neural networks and prevalent machine learning algorithms. The experimental results on benchmark datasets widely used in machine learning showed overall strong and weak points of the GFMM classifier. These outcomes also informed potential research directions for this class of machine learning algorithms in the future., Comment: 18 pages, 7 figures, 12 tables

Recent advances in tracking sensors and pose estimation software enable smart systems to use trajectories of skeleton joint locations for supervised learning. We study the problem of accurately recognizing sign language words, which is key to narrowing the communication gap between hard and non-hard of hearing people. Our method explores a geometric feature space that we call `sub-skeleton' aspects of movement. We assess similarity of feature space trajectories using natural, speed invariant distance measures, which enables clear and insightful nearest neighbor classification. The simplicity and scalability of our basic method allows for immediate application in different data domains with little to no parameter tuning. We demonstrate the effectiveness of our basic method, and a boosted variation, with experiments on data from different application domains and tracking technologies. Surprisingly, our simple methods improve sign recognition over recent, state-of-the-art approaches., To appear in Proc. of the 27th International Conference on Database Systems for Advanced Applications (DASFAA-2022)

The ACM Multimedia 2022 Computational Paralinguistics Challenge addresses four different problems for the first time in a research competition under well-defined conditions: In the Vocalisations and Stuttering Sub-Challenges, a classification on human non-verbal vocalisations and speech has to be made; the Activity Sub-Challenge aims at beyond-audio human activity recognition from smartwatch sensor data; and in the Mosquitoes Sub-Challenge, mosquitoes need to be detected. We describe the Sub-Challenges, baseline feature extraction, and classifiers based on the usual ComPaRE and BoAW features, the auDeep toolkit, and deep feature extraction from pre-trained CNNs using the DeepSpectRum toolkit; in addition, we add end-to-end sequential modelling, and a log-mel-128-BNN., Comment: 5 pages, part of the ACM Multimedia 2022 Grand Challenge "The ACM Multimedia 2022 Computational Paralinguistics Challenge (ComParE 2022)"

Early detection of breast cancer in mammography screening via deep-learning based computer-aided detection systems shows promising potential in improving the curability and mortality rates of breast cancer. However, many clinical centres are restricted in the amount and heterogeneity of available data to train such models to (i) achieve promising performance and to (ii) generalise well across acquisition protocols and domains. As sharing data between centres is restricted due to patient privacy concerns, we propose a potential solution: sharing trained generative models between centres as substitute for real patient data. In this work, we use three well known mammography datasets to simulate three different centres, where one centre receives the trained generator of Generative Adversarial Networks (GANs) from the two remaining centres in order to augment the size and heterogeneity of its training dataset. We evaluate the utility of this approach on mammography patch classification on the test set of the GAN-receiving centre using two different classification models, (a) a convolutional neural network and (b) a transformer neural network. Our experiments demonstrate that shared GANs notably increase the performance of both transformer and convolutional classification models and highlight this approach as a viable alternative to inter-centre data sharing., Comment: Draft accepted as oral presentation at International Workshop on Breast Imaging (IWBI) 2022. 9 pages, 3 figures

Machine learning (ML) offers powerful methods for detecting and modeling associations often in data with large feature spaces and complex associations. Many useful tools/packages (e.g. scikit-learn) have been developed to make the various elements of data handling, processing, modeling, and interpretation accessible. However, it is not trivial for most investigators to assemble these elements into a rigorous, replicatable, unbiased, and effective data analysis pipeline. Automated machine learning (AutoML) seeks to address these issues by simplifying the process of ML analysis for all. Here, we introduce STREAMLINE, a simple, transparent, end-to-end AutoML pipeline designed as a framework to easily conduct rigorous ML modeling and analysis (limited initially to binary classification). STREAMLINE is specifically designed to compare performance between datasets, ML algorithms, and other AutoML tools. It is unique among other autoML tools by offering a fully transparent and consistent baseline of comparison using a carefully designed series of pipeline elements including: (1) exploratory analysis, (2) basic data cleaning, (3) cross validation partitioning, (4) data scaling and imputation, (5) filter-based feature importance estimation, (6) collective feature selection, (7) ML modeling with `Optuna' hyperparameter optimization across 15 established algorithms (including less well-known Genetic Programming and rule-based ML), (8) evaluation across 16 classification metrics, (9) model feature importance estimation, (10) statistical significance comparisons, and (11) automatically exporting all results, plots, a PDF summary report, and models that can be easily applied to replication data., Comment: 24 pages, 15 figures, submitted for publication in Genetic Programming Theory and Practice Proceedings

In order to perform GP predictions fast in large geospatial fields with small-scale variations, a computational complexity that is independent of the number of measurements $N$ and the size of the field is crucial. In this setting, GP approximations using $m$ basis functions requires $\mathcal{O}(Nm^2+m^3)$ computations. Using finite-support basis functions reduces the required number of computations to perform a single prediction to $\mathcal{O}(m^3)$, after a one-time training cost of $O(N)$. The prediction cost increases with increasing field size, as the number of required basis functions $m$ grows with the size of the field relative to the size of the spatial variations. To prevent the prediction speed from depending on field size, we propose leveraging the property that a subset of the trained system is a trained subset of the system to use only a local subset of $m'\ll m$ finite-support basis functions centered around each prediction point to perform predictions. Our proposed approximation requires $\mathcal{O}(m'^3)$ operations to perform each prediction after a one-time training cost of $\mathcal{O}(N)$. We show on real-life spatial data that our approach matches the prediction error of state-of-the-art methods and that it performs faster predictions, also compared to state-of-the-art approximations that lower the prediction cost of $\mathcal{O}(m^3)$ to $\mathcal{O}(m\log(m))$ using a conjugate gradient solver. Finally, we demonstrate that our approach can perform fast predictions on a global bathymetry dataset using millions of basis functions and tens of millions of measurements on a laptop computer.

Plotting a learner's generalization performance against the training set size results in a so-called learning curve. This tool, providing insight in the behavior of the learner, is also practically valuable for model selection, predicting the effect of more training data, and reducing the computational complexity of training. We set out to make the (ideal) learning curve concept precise and briefly discuss the aforementioned usages of such curves. The larger part of this survey's focus, however, is on learning curves that show that more data does not necessarily leads to better generalization performance. A result that seems surprising to many researchers in the field of artificial intelligence. We point out the significance of these findings and conclude our survey with an overview and discussion of open problems in this area that warrant further theoretical and empirical investigation., Comment: arXiv admin note: substantial text overlap with arXiv:2103.10948

A keyword search on constrained clustering on Web-of-Science returned just under 3,000 documents. We ran automatic analyses of those, and compiled our own bibliography of 183 papers which we analysed in more detail based on their topic and experimental study, if any. This paper presents general trends of the area and its sub-topics by Pareto analysis, using citation count and year of publication. We list available software and analyse the experimental sections of our reference collection. We found a notable lack of large comparison experiments. Among the topics we reviewed, applications studies were most abundant recently, alongside deep learning, active learning and ensemble learning., Comment: 18 pages, 11 figures, 177 references

Whilst diffusion probabilistic models can generate high quality image content, key limitations remain in terms of both generating high-resolution imagery and their associated high computational requirements. Recent Vector-Quantized image models have overcome this limitation of image resolution but are prohibitively slow and unidirectional as they generate tokens via element-wise autoregressive sampling from the prior. By contrast, in this paper we propose a novel discrete diffusion probabilistic model prior which enables parallel prediction of Vector-Quantized tokens by using an unconstrained Transformer architecture as the backbone. During training, tokens are randomly masked in an order-agnostic manner and the Transformer learns to predict the original tokens. This parallelism of Vector-Quantized token prediction in turn facilitates unconditional generation of globally consistent high-resolution and diverse imagery at a fraction of the computational expense. In this manner, we can generate image resolutions exceeding that of the original training set samples whilst additionally provisioning per-image likelihood estimates (in a departure from generative adversarial approaches). Our approach achieves state-of-the-art results in terms of Density (LSUN Bedroom: 1.51; LSUN Churches: 1.12; FFHQ: 1.20) and Coverage (LSUN Bedroom: 0.83; LSUN Churches: 0.73; FFHQ: 0.80), and performs competitively on FID (LSUN Bedroom: 3.64; LSUN Churches: 4.07; FFHQ: 6.11) whilst offering advantages in terms of both computation and reduced training set requirements., 19 pages, 14 figures

Datasets can be biased due to societal inequities, human biases, under-representation of minorities, etc. Our goal is to certify that models produced by a learning algorithm are pointwise-robust to potential dataset biases. This is a challenging problem: it entails learning models for a large, or even infinite, number of datasets, ensuring that they all produce the same prediction. We focus on decision-tree learning due to the interpretable nature of the models. Our approach allows programmatically specifying bias models across a variety of dimensions (e.g., missing data for minorities), composing types of bias, and targeting bias towards a specific group. To certify robustness, we use a novel symbolic technique to evaluate a decision-tree learner on a large, or infinite, number of datasets, certifying that each and every dataset produces the same prediction for a specific test point. We evaluate our approach on datasets that are commonly used in the fairness literature, and demonstrate our approach's viability on a range of bias models., To be published at NeurIPS 2021. 22 pages, 4 figures

Several approaches have been proposed in recent literature to alleviate the long-tail problem, mainly in object classification tasks. In this paper, we make the first large-scale study concerning the task of Long-Tail Visual Relationship Recognition (LTVRR). LTVRR aims at improving the learning of structured visual relationships that come from the long-tail (e.g., "rabbit grazing on grass"). In this setup, the subject, relation, and object classes each follow a long-tail distribution. To begin our study and make a future benchmark for the community, we introduce two LTVRR-related benchmarks, dubbed VG8K-LT and GQA-LT, built upon the widely used Visual Genome and GQA datasets. We use these benchmarks to study the performance of several state-of-the-art long-tail models on the LTVRR setup. Lastly, we propose a visiolinguistic hubless (VilHub) loss and a Mixup augmentation technique adapted to LTVRR setup, dubbed as RelMix. Both VilHub and RelMix can be easily integrated on top of existing models and despite being simple, our results show that they can remarkably improve the performance, especially on tail classes. Benchmarks, code, and models have been made available at: https://github.com/Vision-CAIR/LTVRR.

This paper proposes a simple yet powerful ensemble classifier, called Random Hyperboxes, constructed from individual hyperbox-based classifiers trained on the random subsets of sample and feature spaces of the training set. We also show a generalization error bound of the proposed classifier based on the strength of the individual hyperbox-based classifiers as well as the correlation among them. The effectiveness of the proposed classifier is analyzed using a carefully selected illustrative example and compared empirically with other popular single and ensemble classifiers via 20 datasets using statistical testing methods. The experimental results confirmed that our proposed method outperformed other fuzzy min-max neural networks, popular learning algorithms, and is competitive with other ensemble methods. Finally, we identify the existing issues related to the generalization error bounds of the real datasets and inform the potential research directions., Comment: @20XX IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works

Providing early diagnosis of cerebral palsy (CP) is key to enhancing the developmental outcomes for those affected. Diagnostic tools such as the General Movements Assessment (GMA), have produced promising results in early diagnosis, however these manual methods can be laborious. In this paper, we propose a new framework for the automated classification of infant body movements, based upon the GMA, which unlike previous methods, also incorporates a visualization framework to aid with interpretability. Our proposed framework segments extracted features to detect the presence of Fidgety Movements (FMs) associated with the GMA spatiotemporally. These features are then used to identify the body-parts with the greatest contribution towards a classification decision and highlight the related body-part segment providing visual feedback to the user. We quantitatively compare the proposed framework's classification performance with several other methods from the literature and qualitatively evaluate the visualization's veracity. Our experimental results show that the proposed method performs more robustly than comparable techniques in this setting whilst simultaneously providing relevant visual interpretability., Comment: Proceedings of the 2021 IEEE EMBS International Conference on Biomedical & Health Informatics (BHI), accepted, 2021

Since the Message Passing (Graph) Neural Networks (MPNNs) have a linear complexity with respect to the number of nodes when applied to sparse graphs, they have been widely implemented and still raise a lot of interest even though their theoretical expressive power is limited to the first order Weisfeiler-Lehman test (1-WL). In this paper, we show that if the graph convolution supports are designed in spectral-domain by a non-linear custom function of eigenvalues and masked with an arbitrary large receptive field, the MPNN is theoretically more powerful than the 1-WL test and experimentally as powerful as a 3-WL existing models, while remaining spatially localized. Moreover, by designing custom filter functions, outputs can have various frequency components that allow the convolution process to learn different relationships between a given input graph signal and its associated properties. So far, the best 3-WL equivalent graph neural networks have a computational complexity in $\mathcal{O}(n^3)$ with memory usage in $\mathcal{O}(n^2)$, consider non-local update mechanism and do not provide the spectral richness of output profile. The proposed method overcomes all these aforementioned problems and reaches state-of-the-art results in many downstream tasks., Comment: 18 pages, 6 figures

Voronoi tessellations are used to partition the Euclidean space into polyhedral regions, which are called Voronoi cells. Labeling the Voronoi cells with the class information, we can map any classification problem into a Voronoi tessellation. In this way, the classification problem changes into a query of just finding the enclosing Voronoi cell. In order to accomplish this task, we have developed a new algorithm which generates a labeled Voronoi tessellation that partitions the training data into polyhedral regions and obtains interclass boundaries as an indirect result. It is called Supervised k-Voxels or in short Super-k. We are introducing Super-k as a foundational new algorithm and opening the possibility of a new family of algorithms. In this paper, it is shown via comparisons on certain datasets that the Super-k algorithm has the potential of providing comparable performance of the well-known SVM family of algorithms with less complexity., 11 pages, 15 figures, "for the source code, see https://github.com/ituamg/super-k"

Modern machine learning systems need to be able to cope with constantly arriving and changing data. Two main areas of research dealing with such scenarios are continual learning and data stream mining. Continual learning focuses on accumulating knowledge and avoiding forgetting, assuming information once learned should be stored. Data stream mining focuses on adaptation to concept drift and discarding outdated information, assuming that only the most recent data is relevant. While these two areas are mainly being developed in separation, they offer complementary views on the problem of learning from dynamic data. There is a need for unifying them, by offering architectures capable of both learning and storing new information, as well as revisiting and adapting to changes in previously seen concepts. We propose a novel continual learning approach that can handle both tasks. Our experience replay method is fueled by a centroid-driven memory storing diverse instances of incrementally arriving classes. This is enhanced with a reactive subspace buffer that tracks concept drift occurrences in previously seen classes and adapts clusters accordingly. The proposed architecture is thus capable of both remembering valid and forgetting outdated information, offering a holistic framework for continual learning under concept drift., IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2021

This paper contributes multivariate versions of seven commonly used elastic similarity and distance measures for time series data analytics. Elastic similarity and distance measures are a class of similarity measures that can compensate for misalignments in the time axis of time series data. We adapt two existing strategies used in a multivariate version of the well-known Dynamic Time Warping (DTW), namely, Independent and Dependent DTW, to these seven measures. While these measures can be applied to various time series analysis tasks, we demonstrate their utility on multivariate time series classification using the nearest neighbor classifier. On 23 well-known datasets, we demonstrate that each of the measures but one achieves the highest accuracy relative to others on at least one dataset, supporting the value of developing a suite of multivariate similarity and distance measures. We also demonstrate that there are datasets for which either the dependent versions of all measures are more accurate than their independent counterparts or vice versa. In addition, we also construct a nearest neighbor-based ensemble of the measures and show that it is competitive to other state-of-the-art single-strategy multivariate time series classifiers., 40 pages, 12 figures

Standard few-shot experiments involve learning to efficiently match previously unseen samples by class. We claim that few-shot learning should be long term, assimilating knowledge for the future, without forgetting previous concepts. In the mammalian brain, the hippocampus is understood to play a significant role in this process, by learning rapidly and consolidating knowledge to the neocortex incrementally over a short period. In this research we tested whether an artificial hippocampal algorithm (AHA), could be used with a conventional Machine Learning (ML) model that learns incrementally analogous to the neocortex, to achieve one-shot learning both short and long term. The results demonstrated that with the addition of AHA, the system could learn in one-shot and consolidate the knowledge for the long term without catastrophic forgetting. This study is one of the first examples of using a CLS model of hippocampus to consolidate memories, and it constitutes a step toward few-shot continual learning., Accepted to 'The International Joint Conference on Neural Networks (IJCNN) 2021' https://www.ijcnn.org/

Preserving privacy is a growing concern in our society where sensors and cameras are ubiquitous. In this work, for the first time, we propose a trainable image acquisition method that removes the sensitive identity revealing information in the optical domain before it reaches the image sensor. The method benefits from a trainable optical convolution kernel which transmits the desired information while filters out the sensitive content. As the sensitive content is suppressed before it reaches the image sensor, it does not enter the digital domain therefore is unretrievable by any sort of privacy attack. This is in contrast with the current digital privacy-preserving methods that are all vulnerable to direct access attack. Also, in contrast with the previous optical privacy-preserving methods that cannot be trained, our method is data-driven and optimized for the specific application at hand. Moreover, there is no additional computation, memory, or power burden on the acquisition system since this processing happens passively in the optical domain and can even be used together and on top of the fully digital privacy-preserving systems. The proposed approach is adaptable to different digital neural networks and content. We demonstrate it for several scenarios such as smile detection as the desired attribute while the gender is filtered out as the sensitive content. We trained the optical kernel in conjunction with two adversarial neural networks where the analysis network tries to detect the desired attribute and the adversarial network tries to detect the sensitive content. We show that this method can reduce 65.1% of sensitive content when it is selected to be the gender and it only loses 7.3% of the desired content. Moreover, we reconstruct the original faces using the deep reconstruction method that confirms the ineffectiveness of reconstruction attacks to obtain the sensitive content., Comment: 9 pages, 9 figures

Continual learning from data streams is among the most important topics in contemporary machine learning. One of the biggest challenges in this domain lies in creating algorithms that can continuously adapt to arriving data. However, previously learned knowledge may become outdated, as streams evolve over time. This phenomenon is known as concept drift and must be detected to facilitate efficient adaptation of the learning model. While there exists a plethora of drift detectors, all of them assume that we are dealing with roughly balanced classes. In the case of imbalanced data streams, those detectors will be biased towards the majority classes, ignoring changes happening in the minority ones. Furthermore, class imbalance may evolve over time and classes may change their roles (majority becoming minority and vice versa). This is especially challenging in the multi-class setting, where relationships among classes become complex. In this paper, we propose a detailed taxonomy of challenges posed by concept drift in multi-class imbalanced data streams, as well as a novel trainable concept drift detector based on Restricted Boltzmann Machine. It is capable of monitoring multiple classes at once and using reconstruction error to detect changes in each of them independently. Our detector utilizes a skew-insensitive loss function that allows it to handle multiple imbalanced distributions. Due to its trainable nature, it is capable of following changes in a stream and evolving class roles, as well as it can deal with local concept drift occurring in minority classes. Extensive experimental study on multi-class drifting data streams, enriched with a detailed analysis of the impact of local drifts and changing imbalance ratios, confirms the high efficacy of our approach., Comment: 37th IEEE International Conference on Data Engineering (ICDE), 2021. arXiv admin note: text overlap with arXiv:2009.09497

Vision-based reinforcement learning (RL) is a promising technique to solve control tasks involving images as the main observation. State-of-the-art RL algorithms still struggle in terms of sample efficiency, especially when using image observations. This has led to an increased attention on integrating state representation learning (SRL) techniques into the RL pipeline. Work in this field demonstrates a substantial improvement in sample efficiency among other benefits. However, to take full advantage of this paradigm, the quality of samples used for training plays a crucial role. More importantly, the diversity of these samples could affect the sample efficiency of vision-based RL, but also its generalization capability. In this work, we present an approach to improve the sample diversity. Our method enhances the exploration capability of the RL algorithms by taking advantage of the SRL setup. Our experiments show that the presented approach outperforms the baseline for all tested environments. These results are most apparent for environments where the baseline method struggles. Even in simple environments, our method stabilizes the training, reduces the reward variance and boosts sample efficiency., Comment: ICRA workshop on Curious Robots 2021

Most artificial neural networks used for object detection and recognition are trained in a fully supervised setup. This is not only very resource consuming as it requires large data sets of labeled examples but also very different from how humans learn. We introduce a setup in which an artificial agent first learns in a simulated world through self-supervised exploration. Following this, the representations learned through interaction with the world can be used to associate semantic concepts such as different types of doors. To do this, we use a method we call fast concept mapping which uses correlated firing patterns of neurons to define and detect semantic concepts. This association works instantaneous with very few labeled examples, similar to what we observe in humans in a phenomenon called fast mapping. Strikingly, this method already identifies objects with as little as one labeled example which highlights the quality of the encoding learned self-supervised through embodiment using curiosity-driven exploration. It therefor presents a feasible strategy for learning concepts without much supervision and shows that through pure interaction with the world meaningful representations of an environment can be learned., Comment: 10 pages. Find associated code and data here: https://github.com/vkakerbeck/FastConceptMapping

Generating realistic sequences is a central task in many machine learning applications. There has been considerable recent progress on building deep generative models for sequence generation tasks. However, the issue of mode-collapsing remains a main issue for the current models. In this paper we propose a GAN-based generic framework to address the problem of mode-collapse in a principled approach. We change the standard GAN objective to maximize a variational lower-bound of the log-likelihood while minimizing the Jensen-Shanon divergence between data and model distributions. We experiment our model with text generation task and show that it can generate realistic text with high diversity., Comment: Accepted in the Third Workshop on Bayesian Deep Learning (NIPS / NeurIPS 2018)

Standard bias and variance (B&V) terminologies were originally defined for the regression setting and their extensions to classification have led to several different models / definitions in the literature. In this paper, we aim to provide the link between the commonly used frameworks of Tumer & Ghosh (T&G) and James. By unifying the two approaches, we relate the B&V defined for the 0/1 loss to the standard B&V of the boundary distributions given for the squared error loss. The closed form relationships provide a deeper understanding of classification performance, and their use is demonstrated in two case studies., 17 pages, 3 figures

The INTERSPEECH 2021 Computational Paralinguistics Challenge addresses four different problems for the first time in a research competition under well-defined conditions: In the COVID-19 Cough and COVID-19 Speech Sub-Challenges, a binary classification on COVID-19 infection has to be made based on coughing sounds and speech; in the Escalation SubChallenge, a three-way assessment of the level of escalation in a dialogue is featured; and in the Primates Sub-Challenge, four species vs background need to be classified. We describe the Sub-Challenges, baseline feature extraction, and classifiers based on the 'usual' COMPARE and BoAW features as well as deep unsupervised representation learning using the AuDeep toolkit, and deep feature extraction from pre-trained CNNs using the Deep Spectrum toolkit; in addition, we add deep end-to-end sequential modelling, and partially linguistic analysis., 5 pages

Deep generative models are a class of techniques that train deep neural networks to model the distribution of training samples. Research has fragmented into various interconnected approaches, each of which make trade-offs including run-time, diversity, and architectural restrictions. In particular, this compendium covers energy-based models, variational autoencoders, generative adversarial networks, autoregressive models, normalizing flows, in addition to numerous hybrid approaches. These techniques are compared and contrasted, explaining the premises behind each and how they are interrelated, while reviewing current state-of-the-art advances and implementations., Comment: 20 pages, 9 figures, will appear in IEEE Transactions on Pattern Analysis and Machine Intelligence

Earth observation offers new insight into anthropogenic changes to nature, and how these changes are effecting (and are effected by) the built environment and the real economy. With the global availability of medium-resolution (10-30m) synthetic aperture radar (SAR) Sentinel-1 and multispectral Sentinel-2 imagery, machine learning can be employed to offer these insights at scale, unbiased to the reporting of companies and countries. In this paper, I introduce DeepSentinel, a data pipeline and experimentation framework for producing general-purpose semantic embeddings of paired Sentinel-1 and Sentinel-2 imagery. I document the development of an extensible corpus of labelled and unlabelled imagery for the purposes of sensor fusion research. With this new dataset I develop a set of experiments applying popular self-supervision methods and encoder architectures to a land cover classification problem. Tile2vec spatial encoding with a self-attention enabled ResNet model outperforms deeper ResNet variants as well as pretraining with variational autoencoding and contrastive loss. All supporting and derived data and code are made publicly available., Comment: Proposal presented at NeurIPS 2020 Climate Change AI Workshop and ESA Phi-Week. Copernicus Masters finalist. 14 pages; 4 figures

Vision-based reinforcement learning (RL) is a promising approach to solve control tasks involving images as the main observation. State-of-the-art RL algorithms still struggle in terms of sample efficiency, especially when using image observations. This has led to increased attention on integrating state representation learning (SRL) techniques into the RL pipeline. Work in this field demonstrates a substantial improvement in sample efficiency among other benefits. However, to take full advantage of this paradigm, the quality of samples used for training plays a crucial role. More importantly, the diversity of these samples could affect the sample efficiency of vision-based RL, but also its generalization capability. In this work, we present an approach to improve sample diversity for state representation learning. Our method enhances the exploration capability of RL algorithms, by taking advantage of the SRL setup. Our experiments show that our proposed approach boosts the visitation of problematic states, improves the learned state representation, and outperforms the baselines for all tested environments. These results are most apparent for environments where the baseline methods struggle. Even in simple environments, our method stabilizes the training, reduces the reward variance, and promotes sample efficiency., Comment: arXiv admin note: substantial text overlap with arXiv:2109.13588

Random Fourier features provide a way to tackle large-scale machine learning problems with kernel methods. Their slow Monte Carlo convergence rate has motivated the research of deterministic Fourier features whose approximation error can decrease exponentially in the number of basis functions. However, due to their tensor product extension to multiple dimensions, these methods suffer heavily from the curse of dimensionality, limiting their applicability to one, two or three-dimensional scenarios. In our approach we overcome said curse of dimensionality by exploiting the tensor product structure of deterministic Fourier features, which enables us to represent the model parameters as a low-rank tensor decomposition. We derive a monotonically converging block coordinate descent algorithm with linear complexity in both the sample size and the dimensionality of the inputs for a regularized squared loss function, allowing to learn a parsimonious model in decomposed form using deterministic Fourier features. We demonstrate by means of numerical experiments how our low-rank tensor approach obtains the same performance of the corresponding nonparametric model, consistently outperforming random Fourier features., Comment: 9 pages, 6 figures. Reviewed version after peer-review. To be published in the proceedings of the Thirty-fifth Conference on Neural Information Processing Systems (NeurIPS 2021)

Modern deep neural networks must demonstrate state-of-the-art accuracy while exhibiting low latency and energy consumption. As such, neural architecture search (NAS) algorithms take these two constraints into account when generating a new architecture. However, efficiency metrics such as latency are typically hardware dependent requiring the NAS algorithm to either measure or predict the architecture latency. Measuring the latency of every evaluated architecture adds a significant amount of time to the NAS process. Here we propose Microprocessor A Priori for Latency Estimation MAPLE that does not rely on transfer learning or domain adaptation but instead generalizes to new hardware by incorporating a prior hardware characteristics during training. MAPLE takes advantage of a novel quantitative strategy to characterize the underlying microprocessor by measuring relevant hardware performance metrics, yielding a fine-grained and expressive hardware descriptor. Moreover, the proposed MAPLE benefits from the tightly coupled I/O between the CPU and GPU and their dependency to predict DNN latency on GPUs while measuring microprocessor performance hardware counters from the CPU feeding the GPU hardware. Through this quantitative strategy as the hardware descriptor, MAPLE can generalize to new hardware via a few shot adaptation strategy where with as few as 3 samples it exhibits a 6% improvement over state-of-the-art methods requiring as much as 10 samples. Experimental results showed that, increasing the few shot adaptation samples to 10 improves the accuracy significantly over the state-of-the-art methods by 12%. Furthermore, it was demonstrated that MAPLE exhibiting 8-10% better accuracy, on average, compared to relevant baselines at any number of adaptation samples., Comment: 13 pages, 4 figures

Mining data streams poses a number of challenges, including the continuous and non-stationary nature of data, the massive volume of information to be processed and constraints put on the computational resources. While there is a number of supervised solutions proposed for this problem in the literature, most of them assume that access to the ground truth (in form of class labels) is unlimited and such information can be instantly utilized when updating the learning system. This is far from being realistic, as one must consider the underlying cost of acquiring labels. Therefore, solutions that can reduce the requirements for ground truth in streaming scenarios are required. In this paper, we propose a novel framework for mining drifting data streams on a budget, by combining information coming from active learning and self-labeling. We introduce several strategies that can take advantage of both intelligent instance selection and semi-supervised procedures, while taking into account the potential presence of concept drift. Such a hybrid approach allows for efficient exploration and exploitation of streaming data structures within realistic labeling budgets. Since our framework works as a wrapper, it may be applied with different learning algorithms. Experimental study, carried out on a diverse set of real-world data streams with various types of concept drift, proves the usefulness of the proposed strategies when dealing with highly limited access to class labels. The presented hybrid approach is especially feasible when one cannot increase a budget for labeling or replace an inefficient classifier. We deliver a set of recommendations regarding areas of applicability for our strategies.

Face recognition in unconstrained environments is challenging due to variations in illumination, quality of sensing, motion blur and etc. An individual's face appearance can vary drastically under different conditions creating a gap between train (source) and varying test (target) data. The domain gap could cause decreased performance levels in direct knowledge transfer from source to target. Despite fine-tuning with domain specific data could be an effective solution, collecting and annotating data for all domains is extremely expensive. To this end, we propose a self-supervised domain learning (SSDL) scheme that trains on triplets mined from unlabelled data. A key factor in effective discriminative learning, is selecting informative triplets. Building on most confident predictions, we follow an "easy-to-hard" scheme of alternate triplet mining and self-learning. Comprehensive experiments on four different benchmarks show that SSDL generalizes well on different domains.

In this paper, we address the problem of vision-based obstacle avoidance for robotic manipulators. This topic poses challenges for both perception and motion generation. While most work in the field aims at improving one of those aspects, we provide a unified framework for approaching this problem. The main goal of this framework is to connect perception and motion by identifying the relationship between the visual input and the corresponding motion representation. To this end, we propose a method for learning reactive obstacle avoidance policies. We evaluate our method on goal-reaching tasks for single and multiple obstacles scenarios. We show the ability of the proposed method to efficiently learn stable obstacle avoidance strategies at a high success rate, while maintaining closed-loop responsiveness required for critical applications like human-robot interaction., Accepted for publication at CoRL 2020

Established experimental procedures for one-shot machine learning do not test the ability to learn or remember specific instances of classes, a key feature of animal intelligence. Distinguishing specific instances is necessary for many real-world tasks, such as remembering which cup belongs to you. Generalisation within classes conflicts with the ability to separate instances of classes, making it difficult to achieve both capabilities within a single architecture. We propose an extension to the standard Omniglot classification-generalisation framework that additionally tests the ability to distinguish specific instances after one exposure and introduces noise and occlusion corruption. Learning is defined as an ability to classify as well as recall training samples. Complementary Learning Systems (CLS) is a popular model of mammalian brain regions believed to play a crucial role in learning from a single exposure to a stimulus. We created an artificial neural network implementation of CLS and applied it to the extended Omniglot benchmark. Our unsupervised model demonstrates comparable performance to existing supervised ANNs on the Omniglot classification task (requiring generalisation), without the need for domain-specific inductive biases. On the extended Omniglot instance-recognition task, the same model also demonstrates significantly better performance than a baseline nearest-neighbour approach, given partial occlusion and noise., To appear at AI 2020 (Australasian Joint Conference on Artificial Intelligence - http://www.ajcai2020.net/)

An automatic image segmentation procedure is an inevitable part of many image analyses and computer vision which deeply affect the rest of the system; therefore, a set of interactive segmentation evaluation methods can substantially simplify the system development process. This entry presents the state of the art of quantitative evaluation metrics for color image segmentation methods by performing an analytical and comparative review of the measures. The decision-making process in selecting a suitable evaluation metric is still very serious because each metric tends to favor a different segmentation method for each benchmark dataset. Furthermore, a conceptual comparison of these metrics is provided at a high level of abstraction and is discussed for understanding the quantitative changes in different image segmentation results., 19 pages, 11 figures, 6 tables, 29 equations, book chapter, 2 authors

An important challenge in metric learning is scalability to both size and dimension of input data. Online metric learning algorithms are proposed to address this challenge. Existing methods are commonly based on (Passive Aggressive) PA approach. Hence, they can rapidly process large volumes of data with an adaptive learning rate. However, these algorithms are based on the Hinge loss and so are not robust against outliers and label noise. Also, existing online methods usually assume training triplets or pairwise constraints are exist in advance. However, many datasets in real-world applications are in the form of input data and their associated labels. We address these challenges by formulating the online Distance-Similarity learning problem with the robust Rescaled hinge loss function. The proposed model is rather general and can be applied to any PA-based online Distance-Similarity algorithm. Also, we develop an efficient robust one-pass triplet construction algorithm. Finally, to provide scalability in high dimensional DML environments, the low-rank version of the proposed methods is presented that not only reduces the computational cost significantly but also keeps the predictive performance of the learned metrics. Also, it provides a straightforward extension of our methods for deep Distance-Similarity learning. We conduct several experiments on datasets from various applications. The results confirm that the proposed methods significantly outperform state-of-the-art online DML methods in the presence of label noise and outliers by a large margin., An Online Distance-Similarity learning approach in noisy environment

General fuzzy min-max neural network (GFMMNN) is one of the efficient neuro-fuzzy systems for data classification. However, one of the downsides of its original learning algorithms is the inability to handle and learn from the mixed-attribute data. While categorical features encoding methods can be used with the GFMMNN learning algorithms, they exhibit a lot of shortcomings. Other approaches proposed in the literature are not suitable for on-line learning as they require entire training data available in the learning phase. With the rapid change in the volume and velocity of streaming data in many application areas, it is increasingly required that the constructed models can learn and adapt to the continuous data changes in real-time without the need for their full retraining or access to the historical data. This paper proposes an extended online learning algorithm for the GFMMNN. The proposed method can handle the datasets with both continuous and categorical features. The extensive experiments confirmed superior and stable classification performance of the proposed approach in comparison to other relevant learning algorithms for the GFMM model.