Direct side information learning for zero-shot regression

Zero-shot learning provides models for targets for which instances are not available, commonly called unobserved targets. The availability of target side information becomes crucial in this context in order to properly induce models for these targets. The literature is plenty of strategies to cope with this scenario, but specifically designed on the basis of a zero-shot classification scenario, mostly in computer vision and image classification, but they are either not applicable or easily extensible for a zero-shot regression framework for which a continuos value is required to be predicted rather than a label. In fact, there is a considerable lack of methods for zero-shot regression in the literature. Two approaches for zero-shot regression that work in a two-phase procedure were recently proposed. They first learn the observed target models through a classical regression learning ignoring the target side information. Then, they aggregate those observed target models afterwards exploiting the target side information and the models for the unobserved targets are induced. Despite both have shown quite good performance because of the different treatment they grant to the common features and to the side information, they exploit features and side information separately, avoiding a global optimization for providing the unobserved target models. The proposal of this paper is a novel method that jointly takes features and side information in a one-phase learning process, but treating side information properly and in a more deserving way than as common features. A specific kernel that properly merges features and side information is proposed for this purpose resulting in a novel approach that exhibits better performance over both artificial and real datasets.

Target inductive methods for zero-shot regression

This research arises from the need to predict the amount of air pollutants in meteorological stations. Air pollution depends on the location of the stations (weather conditions and activities in the surroundings). Frequently, the surrounding information is not considered in the learning process. This information is known beforehand in the absence of unobserved weather conditions and remains constant for the same station. Considering the surrounding information as side information facilitates the generalization for predicting pollutants in new stations, leading to a zero-shot regression scenario. Available methods in zero-shot typically lean towards classification, and are not easily extensible to regression. This paper proposes two zero-shot methods for regression. The first method is a similarity based approach that learns models from features and aggregates them using side information. However, potential knowledge of the feature models may be lost in the aggregation. The second method overcomes this drawback by replacing the aggregation procedure and learning the correspondence between side information and feature-induced models, instead. Both proposals are compared with a baseline procedure using artificial datasets, UCI repository communities and crime datasets, and the pollutants. Both approaches outperform the baseline method, but the parameter learning approach manifests its superiority over the similarity based method.

Regularized boosting with an increasing coefficient magnitude stop criterion as meta-learner in hyperparameter optimization stacking ensemble

In Hyperparameter Optimization (HPO), only the hyperparameter configuration with the best performance is chosen after performing several trials, then, discarding the effort of training all the models with every hyperparameter configuration trial and performing an ensemble of all them. This ensemble consists of simply averaging the model predictions or weighting the models by a certain probability. Recently, other more sophisticated ensemble strategies, such as the Caruana method or the stacking strategy has been proposed. On the one hand, the Caruana method performs well in HPO ensemble, since it is not affected by the effects of multicollinearity, which is prevalent in HPO. It just computes the average over a subset of predictions with replacement. But it does not benefit from the generalization power of a learning process. On the other hand, stacking methods include a learning procedure since a meta-learner is required to perform the ensemble. Yet, one hardly finds advice about which meta-learner is adequate. Besides, some meta-learners may suffer from the effects of multicollinearity or need to be tuned to reduce them. This paper explores meta-learners for stacking ensemble in HPO, free of hyperparameter tuning, able to reduce the effects of multicollinearity and considering the ensemble learning process generalization power. At this respect, the boosting strategy seems promising as a stacking meta-learner. In fact, it completely removes the effects of multicollinearity. This paper also proposes an implicit regularization in the classical boosting method and a novel non-parametric stop criterion suitable only for boosting and specifically designed for HPO. The synergy between these two improvements over boosting exhibits competitive and promising predictive power performance compared to other existing meta-learners and ensemble approaches for HPO other than the stacking ensemble.

Capturing waste collection planning expert knowledge in a fitness function through preference learning

This paper copes with the COGERSA waste collection process. Up to now, experts have been manually designed the process using a trial and error mechanism. This process is not globally optimized, since it has been progressively and locally built as council demands appear. Planning optimization algorithms usually solve it, but they need a fitness function to evaluate a route planning quality. The drawback is that even experts are not able to propose one in a straightforward way due to the complexity of the process. Hence, the goal of this paper is to build a fitness function though a preference framework, taking advantage of the available expert knowledge and expertise. Several key performance indicators together with preference judgments are carefully established according to the experts for learning a promising fitness function. Particularly, the additivity property of them makes the task be much more affordable, since it allows to work with routes rather than with route plannings. Besides, a feature selection analysis is performed over such indicators, since the experts suspect of a potential existing (but unknown) redundancy among them. The experiment results confirm this hypothesis, since the best $C-$index ($98\%$ against around $94\%$) is reached when 6 or 8 out of 21 indicators are taken. Particularly, truck load seems to be a highly promising key performance indicator, together to the travelled distance along non-main roads. A comparison with other existing approaches shows that the proposed method clearly outperforms them, since the $C-$index goes from $72\%$ or $90\%$ to $98\%$.

Improving importance estimation in covariate shift for providing accurate prediction error

In traditional Machine Learning, the algorithms predictions are based on the assumption that the data follows the same distribution in both the training and the test datasets. However, in real world data this condition does not hold and, for instance, the distribution of the covariates changes whereas the conditional distribution of the targets remains unchanged. This situation is called covariate shift problem where standard error estimation may be no longer accurate. In this context, the importance is a measure commonly used to alleviate the influence of covariate shift on error estimations. The main drawback is that it is not easy to compute. The Kullback-Leibler Importance Estimation Procedure (KLIEP) is capable of estimating importance in a promising way. Despite its good performance, it fails to ignore target information, since it only includes the covariates information for computing the importance. In this direction, this paper explores the potential performance improvement if target information is considered in the computation of the importance. Then, a redefinition of the importance arises in order to be generalized in this way. Besides the potential improvement in performance, including target information make possible the application to a real application about plankton classification that motivates this research and characterized by its great dimensionality, since considering targets rather than covariates reduces the computation and the noise in the covariates. The impact of taking target information is also explored when Logistic Regression (LR), Kernel Mean Matching (KMM), Ensemble Kernel Mean Matching (EKMM) and the naive predecessor of KLIEP called Kernel Density Estimation (KDE) methods estimate the importance. The experimental results lead to a more accurate error estimation using target information, especially in case of the more promising method KLIEP.