Carlos Soares

Luis Roque, Carlos Soares, Luís Torgo

We introduce the Robustness of Hierarchically Organized Time Series (RHiOTS) framework, designed to assess the robustness of hierarchical time series forecasting models and algorithms on real-world datasets. Hierarchical time series, where lower-level forecasts must sum to upper-level ones, are prevalent in various contexts, such as retail sales across countries. Current empirical evaluations of forecasting methods are often limited to a small set of benchmark datasets, offering a narrow view of algorithm behavior. RHiOTS addresses this gap by systematically altering existing datasets and modifying the characteristics of individual series and their interrelations. It uses a set of parameterizable transformations to simulate those changes in the data distribution. Additionally, RHiOTS incorporates an innovative visualization component, turning complex, multidimensional robustness evaluation results into intuitive, easily interpretable visuals. This approach allows an in-depth analysis of algorithm and model behavior under diverse conditions. We illustrate the use of RHiOTS by analyzing the predictive performance of several algorithms. Our findings show that traditional statistical methods are more robust than state-of-the-art deep learning algorithms, except when the transformation effect is highly disruptive. Furthermore, we found no significant differences in the robustness of the algorithms when applying specific reconciliation methods, such as MinT. RHiOTS provides researchers with a comprehensive tool for understanding the nuanced behavior of forecasting algorithms, offering a more reliable basis for selecting the most appropriate method for a given problem.

João Lucas Martins, João Bravo, Ana Sofia Gomes et al.

Financial fraud is the cause of multi-billion dollar losses annually. Traditionally, fraud detection systems rely on rules due to their transparency and interpretability, key features in domains where decisions need to be explained. However, rule systems require significant input from domain experts to create and tune, an issue that rule induction algorithms attempt to mitigate by inferring rules directly from data. We explore the application of these algorithms to fraud detection, where rule systems are constrained to have a low false positive rate (FPR) or alert rate, by proposing RIFF, a rule induction algorithm that distills a low FPR rule set directly from decision trees. Our experiments show that the induced rules are often able to maintain or improve performance of the original models for low FPR tasks, while substantially reducing their complexity and outperforming rules hand-tuned by experts.

Inês Gomes, Luís F. Teixeira, Jan N. van Rijn et al.

The increasing use of deep learning across various domains highlights the importance of understanding the decision-making processes of these black-box models. Recent research focusing on the decision boundaries of deep classifiers, relies on generated synthetic instances in areas of low confidence, uncovering samples that challenge both models and humans. We propose a novel approach to enhance the interpretability of deep binary classifiers by selecting representative samples from the decision boundary - prototypes - and applying post-model explanation algorithms. We evaluate the effectiveness of our approach through 2D visualizations and GradientSHAP analysis. Our experiments demonstrate the potential of the proposed method, revealing distinct and compact clusters and diverse prototypes that capture essential features that lead to low-confidence decisions. By offering a more aggregated view of deep classifiers' decision boundaries, our work contributes to the responsible development and deployment of reliable machine learning systems.

Luis Amorim, Moises Santos, Paulo J. Azevedo et al.

Data augmentation is a crucial tool in time series forecasting, especially for deep learning architectures that require a large training sample size to generalize effectively. However, extensive datasets are not always available in real-world scenarios. Although many data augmentation methods exist, their limitations include the use of transformations that do not adequately preserve data properties. This paper introduces Grasynda, a novel graph-based approach for synthetic time series generation that: (1) converts univariate time series into a network structure using a graph representation, where each state is a node and each transition is represented as a directed edge; and (2) encodes their temporal dynamics in a transition probability matrix. We performed an extensive evaluation of Grasynda as a data augmentation method for time series forecasting. We use three neural network variations on six benchmark datasets. The results indicate that Grasynda consistently outperforms other time series data augmentation methods, including ones used in state-of-the-art time series foundation models. The method and all experiments are publicly available.

Abolfazl Mohammadi-Seif, Carlos Soares, Rita P. Ribeiro et al.

Despite the strong predictive performance achieved by machine learning models across many application domains, assessing their trustworthiness through reliable estimates of predictive confidence remains a critical challenge. This issue arises in scenarios where the likelihood of error inferred from learned representations follows a bimodal distribution, resulting from the coexistence of confident and ambiguous predictions. Standard regression approaches often struggle to adequately express this predictive uncertainty, as they implicitly assume unimodal Gaussian noise, leading to mean-collapse behavior in such settings. Although Mixture Density Networks (MDNs) can represent different distributions, they suffer from severe optimization instability. We propose a family of distribution-aware loss functions integrating normalized RMSE with Wasserstein and Cramér distances. When applied to standard deep regression models, our approach recovers bimodal distributions without the volatility of mixture models. Validated across four experimental stages, our results show that the proposed Wasserstein loss establishes a new Pareto efficiency frontier: matching the stability of standard regression losses like MSE in unimodal tasks while reducing Jensen-Shannon Divergence by 45% on complex bimodal datasets. Our framework strictly dominates MDNs in both fidelity and robustness, offering a reliable tool for aleatoric uncertainty estimation in trustworthy AI systems.

Vitor Cerqueira, Luis Torgo, Carlos Soares

Time series forecasting is one of the most active research topics. Machine learning methods have been increasingly adopted to solve these predictive tasks. However, in a recent work, these were shown to systematically present a lower predictive performance relative to simple statistical methods. In this work, we counter these results. We show that these are only valid under an extremely low sample size. Using a learning curve method, our results suggest that machine learning methods improve their relative predictive performance as the sample size grows. The code to reproduce the experiments is available at https://github.com/vcerqueira/MLforForecasting.

Pedro Saleiro, Natasa Milic-Frayling, Eduarda Mendes Rodrigues et al.

Improvements of entity-relationship (E-R) search techniques have been hampered by a lack of test collections, particularly for complex queries involving multiple entities and relationships. In this paper we describe a method for generating E-R test queries to support comprehensive E-R search experiments. Queries and relevance judgments are created from content that exists in a tabular form where columns represent entity types and the table structure implies one or more relationships among the entities. Editorial work involves creating natural language queries based on relationships represented by the entries in the table. We have publicly released the RELink test collection comprising 600 queries and relevance judgments obtained from a sample of Wikipedia List-of-lists-of-lists tables. The latter comprise tuples of entities that are extracted from columns and labelled by corresponding entity types and relationships they represent. In order to facilitate research in complex E-R retrieval, we have created and released as open source the RELink Framework that includes Apache Lucene indexing and search specifically tailored to E-R retrieval. RELink includes entity and relationship indexing based on the ClueWeb-09-B Web collection with FACC1 text span annotations linked to Wikipedia entities. With ready to use search resources and a comprehensive test collection, we support community in pursuing E-R research at scale.

Aníbal Silva, Moisés Santos, André Restivo et al.

Tabular data remains a challenging domain for generative models. In particular, the standard Variational Autoencoder (VAE) architecture, typically composed of multilayer perceptrons, struggles to model relationships between features, especially when handling mixed data types. In contrast, Transformers, through their attention mechanism, are better suited for capturing complex feature interactions. In this paper, we empirically investigate the impact of integrating Transformers into different components of a VAE. We conduct experiments on 57 datasets from the OpenML CC18 suite and draw two main conclusions. First, results indicate that positioning Transformers to leverage latent and decoder representations leads to a trade-off between fidelity and diversity. Second, we observe a high similarity between consecutive blocks of a Transformer in all components. In particular, in the decoder, the relationship between the input and output of a Transformer is approximately linear.

Inês Oliveira e Silva, Sérgio Jesus, Iker Perez et al.

Shapley values are a cornerstone of explainable AI, yet their proliferation into competing formulations has created a fragmented landscape with little consensus on practical deployment. While theoretical differences are well-documented, evaluation remains reliant on quantitative proxies whose alignment with human utility is unverified. In this work, we use a unified amortized framework to isolate semantic differences between eight Shapley variants under the low-latency constraints of operational risk workflows. We conduct a large-scale empirical evaluation across four risk datasets and a realistic fraud-detection environment involving professional analysts and 3,735 case reviews. Our results reveal a fundamental misalignment: standard quantitative metrics, such as sparsity and faithfulness, are decoupled from human-perceived clarity and decision utility. Furthermore, while no formulation improved objective analyst performance, explanations consistently increased decision confidence, signaling a critical risk of automation bias in high-stakes settings. These findings suggest that current evaluation proxies are insufficient for predicting downstream human impact, and we provide evidence-based guidance for selecting formulations and metrics in operational decision systems.

José Leites, Vitor Cerqueira, Carlos Soares

Most forecasting methods use recent past observations (lags) to model the future values of univariate time series. Selecting an adequate number of lags is important for training accurate forecasting models. Several approaches and heuristics have been devised to solve this task. However, there is no consensus about what the best approach is. Besides, lag selection procedures have been developed based on local models and classical forecasting techniques such as ARIMA. We bridge this gap in the literature by carrying out an extensive empirical analysis of different lag selection methods. We focus on deep learning methods trained in a global approach, i.e., on datasets comprising multiple univariate time series. The experiments were carried out using three benchmark databases that contain a total of 2411 univariate time series. The results indicate that the lag size is a relevant parameter for accurate forecasts. In particular, excessively small or excessively large lag sizes have a considerable negative impact on forecasting performance. Cross-validation approaches show the best performance for lag selection, but this performance is comparable with simple heuristics.

Luis Roque, Carlos Soares, Vitor Cerqueira et al.

The importance of time series forecasting drives continuous research and the development of new approaches to tackle this problem. Typically, these methods are introduced through empirical studies that frequently claim superior accuracy for the proposed approaches. Nevertheless, concerns are rising about the reliability and generalizability of these results due to limitations in experimental setups. This paper addresses a critical limitation: the number and representativeness of the datasets used. We investigate the impact of dataset selection bias, particularly the practice of cherry-picking datasets, on the performance evaluation of forecasting methods. Through empirical analysis with a diverse set of benchmark datasets, our findings reveal that cherry-picking datasets can significantly distort the perceived performance of methods, often exaggerating their effectiveness. Furthermore, our results demonstrate that by selectively choosing just four datasets - what most studies report - 46% of methods could be deemed best in class, and 77% could rank within the top three. Additionally, recent deep learning-based approaches show high sensitivity to dataset selection, whereas classical methods exhibit greater robustness. Finally, our results indicate that, when empirically validating forecasting algorithms on a subset of the benchmarks, increasing the number of datasets tested from 3 to 6 reduces the risk of incorrectly identifying an algorithm as the best one by approximately 40%. Our study highlights the critical need for comprehensive evaluation frameworks that more accurately reflect real-world scenarios. Adopting such frameworks will ensure the development of robust and reliable forecasting methods.

Vitor Cerqueira, Nuno Moniz, Ricardo Inácio et al.

Recent state-of-the-art forecasting methods are trained on collections of time series. These methods, often referred to as global models, can capture common patterns in different time series to improve their generalization performance. However, they require large amounts of data that might not be readily available. Besides this, global models sometimes fail to capture relevant patterns unique to a particular time series. In these cases, data augmentation can be useful to increase the sample size of time series datasets. The main contribution of this work is a novel method for generating univariate time series synthetic samples. Our approach stems from the insight that the observations concerning a particular time series of interest represent only a small fraction of all observations. In this context, we frame the problem of training a forecasting model as an imbalanced learning task. Oversampling strategies are popular approaches used to deal with the imbalance problem in machine learning. We use these techniques to create synthetic time series observations and improve the accuracy of forecasting models. We carried out experiments using 7 different databases that contain a total of 5502 univariate time series. We found that the proposed solution outperforms both a global and a local model, thus providing a better trade-off between these two approaches.

Vitor Cerqueira, Moisés Santos, Luis Roque et al.

Deep learning approaches are increasingly used to tackle forecasting tasks involving datasets with multiple univariate time series. A key factor in the successful application of these methods is a large enough training sample size, which is not always available. Synthetic data generation techniques can be applied in these scenarios to augment the dataset. Data augmentation is typically applied offline before training a model. However, when training with mini-batches, some batches may contain a disproportionate number of synthetic samples that do not align well with the original data characteristics. This work introduces an online data augmentation framework that generates synthetic samples during the training of neural networks. By creating synthetic samples for each batch alongside their original counterparts, we maintain a balanced representation between real and synthetic data throughout the training process. This approach fits naturally with the iterative nature of neural network training and eliminates the need to store large augmented datasets. We validated the proposed framework using 3797 time series from 6 benchmark datasets, three neural architectures, and seven synthetic data generation techniques. The experiments suggest that online data augmentation leads to better forecasting performance compared to offline data augmentation or no augmentation approaches. The framework and experiments are publicly available.

João Vitorino, Eva Maia, Isabel Praça et al.

Due to the susceptibility of Artificial Intelligence (AI) to data perturbations and adversarial examples, it is crucial to perform a thorough robustness evaluation before any Machine Learning (ML) model is deployed. However, examining a model's decision boundaries and identifying potential vulnerabilities typically requires access to the training and testing datasets, which may pose risks to data privacy and confidentiality. To improve transparency in organizations that handle confidential data or manage critical infrastructure, it is essential to allow external verification and validation of AI without the disclosure of private datasets. This paper presents Systematic Pattern Analysis (SPATA), a deterministic method that converts any tabular dataset to a domain-independent representation of its statistical patterns, to provide more detailed and transparent data cards. SPATA computes the projection of each data instance into a discrete space where they can be analyzed and compared, without risking data leakage. These projected datasets can be reliably used for the evaluation of how different features affect ML model robustness and for the generation of interpretable explanations of their behavior, contributing to more trustworthy AI.

Luis Roque, Carlos Soares, Vitor Cerqueira et al.

Data augmentation is gaining importance across various aspects of time series analysis, from forecasting to classification and anomaly detection tasks. We introduce the Latent Generative Transformer Augmentation (L-GTA) model, a generative approach using a transformer-based variational recurrent autoencoder. This model uses controlled transformations within the latent space of the model to generate new time series that preserve the intrinsic properties of the original dataset. L-GTA enables the application of diverse transformations, ranging from simple jittering to magnitude warping, and combining these basic transformations to generate more complex synthetic time series datasets. Our evaluation of several real-world datasets demonstrates the ability of L-GTA to produce more reliable, consistent, and controllable augmented data. This translates into significant improvements in predictive accuracy and similarity measures compared to direct transformation methods.

Ricardo Ribeiro Pereira, Jacopo Bono, Hugo Ferreira et al.

When the available data for a target domain is limited, transfer learning (TL) methods can be used to develop models on related data-rich domains, before deploying them on the target domain. However, these TL methods are typically designed with specific, static assumptions on the amount of available labeled and unlabeled target data. This is in contrast with many real world applications, where the availability of data and corresponding labels varies over time. Since the evaluation of the TL methods is typically also performed under the same static data availability assumptions, this would lead to unrealistic expectations concerning their performance in real world settings. To support a more realistic evaluation and comparison of TL algorithms and models, we propose a data manipulation framework that (1) simulates varying data availability scenarios over time, (2) creates multiple domains through resampling of a given dataset and (3) introduces inter-domain variability by applying realistic domain transformations, e.g., creating a variety of potentially time-dependent covariate and concept shifts. These capabilities enable simulation of a large number of realistic variants of the experiments, in turn providing more information about the potential behavior of algorithms when deployed in dynamic settings. We demonstrate the usefulness of the proposed framework by performing a case study on a proprietary real-world suite of card payment datasets. Given the confidential nature of the case study, we also illustrate the use of the framework on the publicly available Bank Account Fraud (BAF) dataset. By providing a methodology for evaluating TL methods over time and in realistic data availability scenarios, our framework facilitates understanding of the behavior of models and algorithms. This leads to better decision making when deploying models for new domains in real-world environments.

Ricardo Inácio, Zafeiris Kokkinogenis, Vitor Cerqueira et al.

Predictive models often reinforce biases which were originally embedded in their training data, through skewed decisions. In such cases, mitigation methods are critical to ensure that, regardless of the prevailing disparities, model outcomes are adjusted to be fair. To assess this, datasets could be systematically generated with specific biases, to train machine learning classifiers. Then, predictive outcomes could aid in the understanding of this bias embedding process. Hence, an agent-based model (ABM), depicting a loan application process that represents various systemic biases across two demographic groups, was developed to produce synthetic datasets. Then, by applying classifiers trained on them to predict loan outcomes, we can assess how biased data leads to unfairness. This highlights a main contribution of this work: a framework for synthetic dataset generation with controllable bias injection. We also contribute with a novel explainability technique, which shows how mitigations affect the way classifiers leverage data features, via second-order Shapley values. In experiments, both offline and online learning approaches are employed. Mitigations are applied at different stages of the modelling pipeline, such as during pre-processing and in-processing.

Rodrigo Tuna, Carlos Soares

Graph generation is an important area in network science. Traditional approaches focus on replicating specific properties of real-world graphs, such as small diameters or power-law degree distributions. Recent advancements in deep learning, particularly with Graph Neural Networks, have enabled data-driven methods to learn and generate graphs without relying on predefined structural properties. Despite these advances, current models are limited by their reliance on node IDs, which restricts their ability to generate graphs larger than the input graph and ignores node attributes. To address these challenges, we propose Latent Graph Sampling Generation (LGSG), a novel framework that leverages diffusion models and node embeddings to generate graphs of varying sizes without retraining. The framework eliminates the dependency on node IDs and captures the distribution of node embeddings and subgraph structures, enabling scalable and flexible graph generation. Experimental results show that LGSG performs on par with baseline models for standard metrics while outperforming them in overlooked ones, such as the tendency of nodes to form clusters. Additionally, it maintains consistent structural characteristics across graphs of different sizes, demonstrating robustness and scalability.

Vitor Cerqueira, Luis Roque, Carlos Soares

Accurate evaluation of forecasting models is essential for ensuring reliable predictions. Current practices for evaluating and comparing forecasting models focus on summarising performance into a single score, using metrics such as SMAPE. While convenient, averaging performance over all samples dilutes relevant information about model behavior under varying conditions. This limitation is especially problematic for time series forecasting, where multiple layers of averaging--across time steps, horizons, and multiple time series in a dataset--can mask relevant performance variations. We address this limitation by proposing ModelRadar, a framework for evaluating univariate time series forecasting models across multiple aspects, such as stationarity, presence of anomalies, or forecasting horizons. We demonstrate the advantages of this framework by comparing 24 forecasting methods, including classical approaches and different machine learning algorithms. NHITS, a state-of-the-art neural network architecture, performs best overall but its superiority varies with forecasting conditions. For instance, concerning the forecasting horizon, we found that NHITS (and also other neural networks) only outperforms classical approaches for multi-step ahead forecasting. Another relevant insight is that classical approaches such as ETS or Theta are notably more robust in the presence of anomalies. These and other findings highlight the importance of aspect-based model evaluation for both practitioners and researchers. ModelRadar is available as a Python package.

Aníbal Silva, André Restivo, Moisés Santos et al.

Generative modeling for tabular data has recently gained significant attention in the Deep Learning domain. Its objective is to estimate the underlying distribution of the data. However, estimating the underlying distribution of tabular data has its unique challenges. Specifically, this data modality is composed of mixed types of features, making it a non-trivial task for a model to learn intra-relationships between them. One approach to address mixture is to embed each feature into a continuous matrix via tokenization, while a solution to capture intra-relationships between variables is via the transformer architecture. In this work, we empirically investigate the potential of using embedding representations on tabular data generation, utilizing tensor contraction layers and transformers to model the underlying distribution of tabular data within Variational Autoencoders. Specifically, we compare four architectural approaches: a baseline VAE model, two variants that focus on tensor contraction layers and transformers respectively, and a hybrid model that integrates both techniques. Our empirical study, conducted across multiple datasets from the OpenML CC18 suite, compares models over density estimation and Machine Learning efficiency metrics. The main takeaway from our results is that leveraging embedding representations with the help of tensor contraction layers improves density estimation metrics, albeit maintaining competitive performance in terms of machine learning efficiency.

Ricardo Inácio, Vitor Cerqueira, Marília Barandas et al.

The effectiveness of univariate forecasting models is often hampered by conditions that cause them stress. A model is considered to be under stress if it shows a negative behaviour, such as higher-than-usual errors or increased uncertainty. Understanding the factors that cause stress to forecasting models is important to improve their reliability, transparency, and utility. This paper addresses this problem by contributing with a novel framework called MAST (Meta-learning and data Augmentation for Stress Testing). The proposed approach aims to model and characterize stress in univariate time series forecasting models, focusing on conditions where they exhibit large errors. In particular, MAST is a meta-learning approach that predicts the probability that a given model will perform poorly on a given time series based on a set of statistical time series features. MAST also encompasses a novel data augmentation technique based on oversampling to improve the metadata concerning stress. We conducted experiments using three benchmark datasets that contain a total of 49.794 time series to validate the performance of MAST. The results suggest that the proposed approach is able to identify conditions that lead to large errors. The method and experiments are publicly available in a repository.

Vitor Cerqueira, Luis Roque, Carlos Soares

Accurate evaluation of forecasting models is essential for ensuring reliable predictions. Current practices for evaluating and comparing forecasting models focus on summarising performance into a single score, using metrics such as SMAPE. We hypothesize that averaging performance over all samples dilutes relevant information about the relative performance of models. Particularly, conditions in which this relative performance is different than the overall accuracy. We address this limitation by proposing a novel framework for evaluating univariate time series forecasting models from multiple perspectives, such as one-step ahead forecasting versus multi-step ahead forecasting. We show the advantages of this framework by comparing a state-of-the-art deep learning approach with classical forecasting techniques. While classical methods (e.g. ARIMA) are long-standing approaches to forecasting, deep neural networks (e.g. NHITS) have recently shown state-of-the-art forecasting performance in benchmark datasets. We conducted extensive experiments that show NHITS generally performs best, but its superiority varies with forecasting conditions. For instance, concerning the forecasting horizon, NHITS only outperforms classical approaches for multi-step ahead forecasting. Another relevant insight is that, when dealing with anomalies, NHITS is outperformed by methods such as Theta. These findings highlight the importance of aspect-based model evaluation.

Rodrigo Tuna, Yassine Baghoussi, Carlos Soares et al.

Forecasting methods are affected by data quality issues in two ways: 1. they are hard to predict, and 2. they may affect the model negatively when it is updated with new data. The latter issue is usually addressed by pre-processing the data to remove those issues. An alternative approach has recently been proposed, Corrector LSTM (cLSTM), which is a Read \& Write Machine Learning (RW-ML) algorithm that changes the data while learning to improve its predictions. Despite promising results being reported, cLSTM is computationally expensive, as it uses a meta-learner to monitor the hidden states of the LSTM. We propose a new RW-ML algorithm, Kernel Corrector LSTM (KcLSTM), that replaces the meta-learner of cLSTM with a simpler method: Kernel Smoothing. We empirically evaluate the forecasting accuracy and the training time of the new algorithm and compare it with cLSTM and LSTM. Results indicate that it is able to decrease the training time while maintaining a competitive forecasting accuracy.

Moisés Santos, André de Carvalho, Carlos Soares

Time series forecasting is a subject of significant scientific and industrial importance. Despite the widespread utilization of forecasting methods, there is a dearth of research aimed at comprehending the conditions under which these methods yield favorable or unfavorable performances. Empirical studies, although common, are challenged by the limited availability of time series datasets, restricting the extraction of reliable insights. To address this limitation, we present tsMorph, a tool for generating semi-synthetic time series through dataset morphing. tsMorph works by creating a sequence of datasets from two original datasets. The characteristics of the generated datasets progressively depart from those of one of the datasets and converge toward the attributes of the other dataset. This method provides a valuable alternative for obtaining substantial datasets. In this paper, we show the benefits of tsMorph by assessing the predictive performance of the Long Short-Term Memory Network and DeepAR forecasting algorithms. The time series used for the experiments comes from the NN5 Competition. The experimental results provide important insights. Notably, the performances of the two algorithms improve proportionally with the frequency of the time series. These experiments confirm that tsMorph can be an effective tool for better understanding the behavior of forecasting algorithms, delivering a pathway to overcoming the limitations posed by empirical studies and enabling more extensive and reliable experiments.

Tarsicio Lucas, Teresa Ludermir, Ricardo Prudencio et al.

The optimization of Artificial Neural Networks (ANNs) is an important task to the success of using these models in real-world applications. The solutions adopted to this task are expensive in general, involving trial-and-error procedures or expert knowledge which are not always available. In this work, we investigated the use of meta-learning to the optimization of ANNs. Meta-learning is a research field aiming to automatically acquiring knowledge which relates features of the learning problems to the performance of the learning algorithms. The meta-learning techniques were originally proposed and evaluated to the algorithm selection problem and after to the optimization of parameters for Support Vector Machines. However, meta-learning can be adopted as a more general strategy to optimize ANN parameters, which motivates new efforts in this research direction. In the current work, we performed a case study using meta-learning to choose the number of hidden nodes for MLP networks, which is an important parameter to be defined aiming a good networks performance. In our work, we generated a base of meta-examples associated to 93 regression problems. Each meta-example was generated from a regression problem and stored: 16 features describing the problem (e.g., number of attributes and correlation among the problem attributes) and the best number of nodes for this problem, empirically chosen from a range of possible values. This set of meta-examples was given as input to a meta-learner which was able to predict the best number of nodes for new problems based on their features. The experiments performed in this case study revealed satisfactory results.

André Baptista, Yassine Baghoussi, Carlos Soares et al.

Forecasting accuracy is reliant on the quality of available past data. Data disruptions can adversely affect the quality of the generated model (e.g. unexpected events such as out-of-stock products when forecasting demand). We address this problem by pastcasting: predicting how data should have been in the past to explain the future better. We propose Pastprop-LSTM, a data-centric backpropagation algorithm that assigns part of the responsibility for errors to the training data and changes it accordingly. We test three variants of Pastprop-LSTM on forecasting competition datasets, M4 and M5, plus the Numenta Anomaly Benchmark. Empirical evaluation indicates that the proposed method can improve forecasting accuracy, especially when the prediction errors of standard LSTM are high. It also demonstrates the potential of the algorithm on datasets containing anomalies.

Vitor Cerqueira, Luis Torgo, Carlos Soares et al.

The predictive advantage of combining several different predictive models is widely accepted. Particularly in time series forecasting problems, this combination is often dynamic to cope with potential non-stationary sources of variation present in the data. Despite their superior predictive performance, ensemble methods entail two main limitations: high computational costs and lack of transparency. These issues often preclude the deployment of such approaches, in favour of simpler yet more efficient and reliable ones. In this paper, we leverage the idea of model compression to address this problem in time series forecasting tasks. Model compression approaches have been mostly unexplored for forecasting. Their application in time series is challenging due to the evolving nature of the data. Further, while the literature focuses on neural networks, we apply model compression to distinct types of methods. In an extensive set of experiments, we show that compressing dynamic forecasting ensembles into an individual model leads to a comparable predictive performance and a drastic reduction in computational costs. Further, the compressed individual model with best average rank is a rule-based regression model. Thus, model compression also leads to benefits in terms of model interpretability. The experiments carried in this paper are fully reproducible.

Vitor Cerqueira, Luis Torgo, Carlos Soares

Evaluating predictive models is a crucial task in predictive analytics. This process is especially challenging with time series data where the observations show temporal dependencies. Several studies have analysed how different performance estimation methods compare with each other for approximating the true loss incurred by a given forecasting model. However, these studies do not address how the estimators behave for model selection: the ability to select the best solution among a set of alternatives. We address this issue and compare a set of estimation methods for model selection in time series forecasting tasks. We attempt to answer two main questions: (i) how often is the best possible model selected by the estimators; and (ii) what is the performance loss when it does not. We empirically found that the accuracy of the estimators for selecting the best solution is low, and the overall forecasting performance loss associated with the model selection process ranges from 1.2% to 2.3%. We also discovered that some factors, such as the sample size, are important in the relative performance of the estimators.

André F. Cruz, Pedro Saleiro, Catarina Belém et al.

Considerable research effort has been guided towards algorithmic fairness but real-world adoption of bias reduction techniques is still scarce. Existing methods are either metric- or model-specific, require access to sensitive attributes at inference time, or carry high development or deployment costs. This work explores the unfairness that emerges when optimizing ML models solely for predictive performance, and how to mitigate it with a simple and easily deployed intervention: fairness-aware hyperparameter optimization (HO). We propose and evaluate fairness-aware variants of three popular HO algorithms: Fair Random Search, Fair TPE, and Fairband. We validate our approach on a real-world bank account opening fraud case-study, as well as on three datasets from the fairness literature. Results show that, without extra training cost, it is feasible to find models with 111% mean fairness increase and just 6% decrease in performance when compared with fairness-blind HO.

Tomas Sousa-Pereira, Tiago Cunha, Carlos Soares

Collaborative Filtering (CF) has become the standard approach to solve recommendation systems (RS) problems. Collaborative Filtering algorithms try to make predictions about interests of a user by collecting the personal interests from multiple users. There are multiple CF algorithms, each one of them with its own biases. It is the Machine Learning practitioner that has to choose the best algorithm for each task beforehand. In Recommender Systems, different algorithms have different performance for different users within the same dataset. Meta Learning (MtL) has been used to choose the best algorithm for a given problem. Meta Learning is usually applied to select algorithms for a whole dataset. Adapting it to select the to the algorithm for a single user in a RS involves several challenges. The most important is the design of the metafeatures which, in typical meta learning, characterize datasets while here, they must characterize a single user. This work presents a new meta-learning based framework named $μ$-cf2vec to select the best algorithm for each user. We propose using Representation Learning techniques to extract the metafeatures. Representation Learning tries to extract representations that can be reused in other learning tasks. In this work we also implement the framework using different RL techniques to evaluate which one can be more useful to solve this task. In the meta level, the meta learning model will use the metafeatures to extract knowledge that will be used to predict the best algorithm for each user. We evaluated an implementation of this framework using MovieLens 20M dataset. Our implementation achieved consistent gains in the meta level, however, in the base level we only achieved marginal gains.

Vitor Cerqueira, Luis Torgo, Carlos Soares

The early detection of anomalous events in time series data is essential in many domains of application. In this paper we deal with critical health events, which represent a significant cause of mortality in intensive care units of hospitals. The timely prediction of these events is crucial for mitigating their consequences and improving healthcare. One of the most common approaches to tackle early anomaly detection problems is standard classification methods. In this paper we propose a novel method that uses a layered learning architecture to address these tasks. One key contribution of our work is the idea of pre-conditional events, which denote arbitrary but computable relaxed versions of the event of interest. We leverage this idea to break the original problem into two hierarchical layers, which we hypothesize are easier to solve. The results suggest that the proposed approach leads to a better performance relative to state of the art approaches for critical health episode prediction.

Vitor Cerqueira, Nuno Moniz, Carlos Soares

Time series forecasting is a challenging task with applications in a wide range of domains. Auto-regression is one of the most common approaches to address these problems. Accordingly, observations are modelled by multiple regression using their past lags as predictor variables. We investigate the extension of auto-regressive processes using statistics which summarise the recent past dynamics of time series. The result of our research is a novel framework called VEST, designed to perform feature engineering using univariate and numeric time series automatically. The proposed approach works in three main steps. First, recent observations are mapped onto different representations. Second, each representation is summarised by statistical functions. Finally, a filter is applied for feature selection. We discovered that combining the features generated by VEST with auto-regression significantly improves forecasting performance. We provide evidence using 90 time series with high sampling frequency. VEST is publicly available online.

André F. Cruz, Pedro Saleiro, Catarina Belém et al.

Considerable research effort has been guided towards algorithmic fairness but there is still no major breakthrough. In practice, an exhaustive search over all possible techniques and hyperparameters is needed to find optimal fairness-accuracy trade-offs. Hence, coupled with the lack of tools for ML practitioners, real-world adoption of bias reduction methods is still scarce. To address this, we present Fairband, a bandit-based fairness-aware hyperparameter optimization (HO) algorithm. Fairband is conceptually simple, resource-efficient, easy to implement, and agnostic to both the objective metrics, model types and the hyperparameter space being explored. Moreover, by introducing fairness notions into HO, we enable seamless and efficient integration of fairness objectives into real-world ML pipelines. We compare Fairband with popular HO methods on four real-world decision-making datasets. We show that Fairband can efficiently navigate the fairness-accuracy trade-off through hyperparameter optimization. Furthermore, without extra training cost, it consistently finds configurations attaining substantially improved fairness at a comparatively small decrease in predictive accuracy.

Cláudio Rebelo de Sá, Paulo Azevedo, Carlos Soares et al.

In this paper we investigate two variants of association rules for preference data, Label Ranking Association Rules and Pairwise Association Rules. Label Ranking Association Rules (LRAR) are the equivalent of Class Association Rules (CAR) for the Label Ranking task. In CAR, the consequent is a single class, to which the example is expected to belong to. In LRAR, the consequent is a ranking of the labels. The generation of LRAR requires special support and confidence measures to assess the similarity of rankings. In this work, we carry out a sensitivity analysis of these similarity-based measures. We want to understand which datasets benefit more from such measures and which parameters have more influence in the accuracy of the model. Furthermore, we propose an alternative type of rules, the Pairwise Association Rules (PAR), which are defined as association rules with a set of pairwise preferences in the consequent. While PAR can be used both as descriptive and predictive models, they are essentially descriptive models. Experimental results show the potential of both approaches.

Pedro Saleiro, Natasa Milic-Frayling, Eduarda Mendes Rodrigues et al.

Entity-Relationship (E-R) Search is a complex case of Entity Search where the goal is to search for multiple unknown entities and relationships connecting them. We assume that a E-R query can be decomposed as a sequence of sub-queries each containing keywords related to a specific entity or relationship. We adopt a probabilistic formulation of the E-R search problem. When creating specific representations for entities (e.g. context terms) and for pairs of entities (i.e. relationships) it is possible to create a graph of probabilistic dependencies between sub-queries and entity plus relationship representations. To the best of our knowledge this represents the first probabilistic model of E-R search. We propose and develop a novel supervised Early Fusion-based model for E-R search, the Entity-Relationship Dependence Model (ERDM). It uses Markov Random Field to model term dependencies of E-R sub-queries and entity/relationship documents. We performed experiments with more than 800M entities and relationships extractions from ClueWeb-09-B with FACC1 entity linking. We obtained promising results using 3 different query collections comprising 469 E-R queries, with results showing that it is possible to perform E-R search without using fix and pre-defined entity and relationship types, enabling a wide range of queries to be addressed.

Tiago Cunha, Carlos Soares, André C. P. L. F. de Carvalho

Algorithm selection using Metalearning aims to find mappings between problem characteristics (i.e. metafeatures) with relative algorithm performance to predict the best algorithm(s) for new datasets. Therefore, it is of the utmost importance that the metafeatures used are informative. In Collaborative Filtering, recent research has created an extensive collection of such metafeatures. However, since these are created based on the practitioner's understanding of the problem, they may not capture the most relevant aspects necessary to properly characterize the problem. We propose to overcome this problem by taking advantage of Representation Learning, which is able to create an alternative problem characterizations by having the data guide the design of the representation instead of the practitioner's opinion. Our hypothesis states that such alternative representations can be used to replace standard metafeatures, hence hence leading to a more robust approach to Metalearning. We propose a novel procedure specially designed for Collaborative Filtering algorithm selection. The procedure models Collaborative Filtering as graphs and extracts distributed representations using graph2vec. Experimental results show that the proposed procedure creates representations that are competitive with state-of-the-art metafeatures, while requiring significantly less data and without virtually any human input.

Adriano Rivolli, Luís P. F. Garcia, Carlos Soares et al.

Meta-learning is increasingly used to support the recommendation of machine learning algorithms and their configurations. Such recommendations are made based on meta-data, consisting of performance evaluations of algorithms on prior datasets, as well as characterizations of these datasets. These characterizations, also called meta-features, describe properties of the data which are predictive for the performance of machine learning algorithms trained on them. Unfortunately, despite being used in a large number of studies, meta-features are not uniformly described, organized and computed, making many empirical studies irreproducible and hard to compare. This paper aims to deal with this by systematizing and standardizing data characterization measures for classification datasets used in meta-learning. Moreover, it presents MFE, a new tool for extracting meta-features from datasets and identifying more subtle reproducibility issues in the literature, proposing guidelines for data characterization that strengthen reproducible empirical research in meta-learning.

Tiago Cunha, Carlos Soares, André C. P. L. F. de Carvalho

To select the best algorithm for a new problem is an expensive and difficult task. However, there are automatic solutions to address this problem: using Metalearning, which takes advantage of problem characteristics (i.e. metafeatures), one is able to predict the relative performance of algorithms. In the Collaborative Filtering scope, recent works have proposed diverse metafeatures describing several dimensions of this problem. Despite interesting and effective findings, it is still unknown whether these are the most effective metafeatures. Hence, this work proposes a new set of graph metafeatures, which approach the Collaborative Filtering problem from a Graph Theory perspective. Furthermore, in order to understand whether metafeatures from multiple dimensions are a better fit, we investigate the effects of comprehensive metafeatures. These metafeatures are a selection of the best metafeatures from all existing Collaborative Filtering metafeatures. The impact of the most representative metafeatures is investigated in a controlled experimental setup. Another contribution we present is the use of a Pareto-Efficient ranking procedure to create multicriteria metatargets. These new rankings of algorithms, which take into account multiple evaluation measures, allow to explore the algorithm selection problem in a fairer and more detailed way. According to the experimental results, the graph metafeatures are a good alternative to related work metafeatures. However, the results have shown that the feature selection procedure used to create the comprehensive metafeatures is is not effective, since there is no gain in predictive performance. Finally, an extensive metaknowledge analysis was conducted to identify the most influential metafeatures.

Arvind Kumar Shekar, Cláudio Rebelo de Sá, Hugo Ferreira et al.

Predicting the health of components in complex dynamic systems such as an automobile poses numerous challenges. The primary aim of such predictive systems is to use the high-dimensional data acquired from different sensors and predict the state-of-health of a particular component, e.g., brake pad. The classical approach involves selecting a smaller set of relevant sensor signals using feature selection and using them to train a machine learning algorithm. However, this fails to address two prominent problems: (1) sensors are susceptible to failure when exposed to extreme conditions over a long periods of time; (2) sensors are electrical devices that can be affected by noise or electrical interference. Using the failed and noisy sensor signals as inputs largely reduce the prediction accuracy. To tackle this problem, it is advantageous to use the information from all sensor signals, so that the failure of one sensor can be compensated by another. In this work, we propose an Artificial Neural Network (ANN) based framework to exploit the information from a large number of signals. Secondly, our framework introduces a data augmentation approach to perform accurate predictions in spite of noisy signals. The plausibility of our framework is validated on real life industrial application from Robert Bosch GmbH.

Tiago Cunha, Carlos Soares, André C. P. L. F. de Carvalho

Automatic solutions which enable the selection of the best algorithms for a new problem are commonly found in the literature. One research area which has recently received considerable efforts is Collaborative Filtering. Existing work includes several approaches using Metalearning, which relate the characteristics of datasets with the performance of the algorithms. This work explores an alternative approach to tackle this problem. Since, in essence, both are recommendation problems, this work uses Collaborative Filtering algorithms to select Collaborative Filtering algorithms. Our approach integrates subsampling landmarkers, which are a data characterization approach commonly used in Metalearning, with a standard Collaborative Filtering method. The experimental results show that CF4CF competes with standard Metalearning strategies in the problem of Collaborative Filtering algorithm selection.

Dylan te Lindert, Cláudio Rebelo de Sá, Carlos Soares et al.

The costs associated with refrigerator equipment often represent more than half of the total energy costs in supermarkets. This presents a good motivation for running these systems efficiently. In this study, we investigate different ways to construct a reference behavior, which can serve as a baseline for judging the performance of energy consumption. We used 3 distinct learning models: Multiple Linear Regression, Random Forests, and Artificial Neural Networks. During our experiments we used a variation of the sliding window method in combination with learning curves. We applied this approach on five different supermarkets, across Portugal. We are able to create baselines using off-the-shelf data mining techniques. Moreover, we found a way to create them based on short term historical data. We believe that our research will serve as a base for future studies, for which we provide interesting directions.

Pedro Saleiro, Luís Sarmento, Eduarda Mendes Rodrigues et al.

This paper describes a preliminary study for producing and distributing a large-scale database of embeddings from the Portuguese Twitter stream. We start by experimenting with a relatively small sample and focusing on three challenges: volume of training data, vocabulary size and intrinsic evaluation metrics. Using a single GPU, we were able to scale up vocabulary size from 2048 words embedded and 500K training examples to 32768 words over 10M training examples while keeping a stable validation loss and approximately linear trend on training time per epoch. We also observed that using less than 50\% of the available training examples for each vocabulary size might result in overfitting. Results on intrinsic evaluation show promising performance for a vocabulary size of 32768 words. Nevertheless, intrinsic evaluation metrics suffer from over-sensitivity to their corresponding cosine similarity thresholds, indicating that a wider range of metrics need to be developed to track progress.

Pedro Saleiro, Natasa Milic-Frayling, Eduarda Mendes Rodrigues et al.

We address the task of entity-relationship (E-R) retrieval, i.e, given a query characterizing types of two or more entities and relationships between them, retrieve the relevant tuples of related entities. Answering E-R queries requires gathering and joining evidence from multiple unstructured documents. In this work, we consider entity and relationships of any type, i.e, characterized by context terms instead of pre-defined types or relationships. We propose a novel IR-centric approach for E-R retrieval, that builds on the basic early fusion design pattern for object retrieval, to provide extensible entity-relationship representations, suitable for complex, multi-relationships queries. We performed experiments with Wikipedia articles as entity representations combined with relationships extracted from ClueWeb-09-B with FACC1 entity linking. We obtained promising results using 3 different query collections comprising 469 E-R queries.

Fábio Pinto, Vítor Cerqueira, Carlos Soares et al.

Machine Learning (ML) has been successfully applied to a wide range of domains and applications. One of the techniques behind most of these successful applications is Ensemble Learning (EL), the field of ML that gave birth to methods such as Random Forests or Boosting. The complexity of applying these techniques together with the market scarcity on ML experts, has created the need for systems that enable a fast and easy drop-in replacement for ML libraries. Automated machine learning (autoML) is the field of ML that attempts to answers these needs. Typically, these systems rely on optimization techniques such as bayesian optimization to lead the search for the best model. Our approach differs from these systems by making use of the most recent advances on metalearning and a learning to rank approach to learn from metadata. We propose autoBagging, an autoML system that automatically ranks 63 bagging workflows by exploiting past performance and dataset characterization. Results on 140 classification datasets from the OpenML platform show that autoBagging can yield better performance than the Average Rank method and achieve results that are not statistically different from an ideal model that systematically selects the best workflow for each dataset. For the purpose of reproducibility and generalizability, autoBagging is publicly available as an R package on CRAN.

Pedro Saleiro, Eduarda Mendes Rodrigues, Carlos Soares et al.

This paper presents the approach developed at the Faculty of Engineering of University of Porto, to participate in SemEval 2017, Task 5: Fine-grained Sentiment Analysis on Financial Microblogs and News. The task consisted in predicting a real continuous variable from -1.0 to +1.0 representing the polarity and intensity of sentiment concerning companies/stocks mentioned in short texts. We modeled the task as a regression analysis problem and combined traditional techniques such as pre-processing short texts, bag-of-words representations and lexical-based features with enhanced financial specific bag-of-embeddings. We used an external collection of tweets and news headlines mentioning companies/stocks from S\&P 500 to create financial word embeddings which are able to capture domain-specific syntactic and semantic similarities. The resulting approach obtained a cosine similarity score of 0.69 in sub-task 5.1 - Microblogs and 0.68 in sub-task 5.2 - News Headlines.

Pedro Saleiro, Jorge Teixeira, Carlos Soares et al.

We present a dynamic web tool that allows interactive search and visualization of large news archives using an entity-centric approach. Users are able to search entities using keyword phrases expressing news stories or events and the system retrieves the most relevant entities to the user query based on automatically extracted and indexed entity profiles. From the computational journalism perspective, TimeMachine allows users to explore media content through time using automatic identification of entity names, jobs, quotations and relations between entities from co-occurrences networks extracted from the news articles. TimeMachine demo is available at http://maquinadotempo.sapo.pt/