Search

We study how web publishers should set their floor prices in order to maximize expected revenues when they have access to two selling mechanisms, namely an ad exchange and header bidding, in order to sell impressions on the real-time bidding market. We consider the publisher’s problem under incomplete information, propose bandit-type algorithms, and show that their regret – the performance loss compared to the optimal algorithm – is sub-linear in the time horizon. Experiments illustrate the effectiveness of our algorithms.

The papers in this special section focus on recent developments and emerging topics in the area of deep fuzzy models that address some of the problems and limitations above. These models have been known under different names, such as hierarchical fuzzy systems and fuzzy networks. They are usually well suited for performing multiple functional compositions at either crisp or linguistic level. Deep learning has gained significant attention within the computational intelligence community in recent years. Its success has been mainly due to the increased power of modern computational platforms in terms of their ability to collect, store, and process large volumes of data. This has led to a substantial increase in the effectiveness and efficiency of data management. As a result, it has become possible to achieve high accuracy for some benchmark learning tasks, such as object classification and image recognition within a short time frame. The most common implementation of deep learning has been through neural networks due to the ability of their layers of neurons to perform multiple functional compositions as part of a multistage learning process.

In many domains that employ machine learning models, both high performing and interpretable models are needed. A typical machine learning task is text classification, where models are hardly interpretable. Topic models, used as topic embeddings, carry the potential to better understand the decisions made by text classification algorithms. With this goal in mind, we propose two new fuzzy topic models; FLSA-W and FLSA-V. Both models are derived from the topic model Fuzzy Latent Semantic Analysis (FLSA). After training each model ten times, we use the mean coherence score to compare the different models with the benchmark models Latent Dirichlet Allocation (LDA) and FLSA. Our proposed models generally lead to higher coherence scores and lower standard deviations than the benchmark models. These proposed models are specifically useful as topic embeddings in text classification, since the coherence scores do not drop for a high number of topics, as opposed to the decay that occurs with LDA and FLSA.

PURPOSE: The focus of bariatric surgery is reduction of weight, reflected in body mass index (BMI). However, the resolution of comorbidity is a second important outcome indicator. The degree of comorbidity is hard to quantify objectively as comorbidities develop gradually and are interdependent. Multiple scoring systems quantifying comorbidity exist but they lack continuity and objectivity. In analogy with BMI as index for weight, the Metabolic Health Index (MHI) is developed as objective quantification of metabolic health status. Laboratory data were used as comorbidities affect biomarkers. Conversely, laboratory data can be used as objectively obtained variables to describe comorbidity.METHODS: Laboratory data were collected and crosschecked by national quality registry entries. Machine learning was applied to develop an ordinal logistic regression model, using 4 clinical and 32 laboratory input variables. The output was mathematically transformed into a continuous score for intuitive interpretation, ranging from 1 to 6 (MHI).RESULTS: In total, 4778 data records of 1595 patients were used. The degree of comorbidity is best described by age at phlebotomy, estimated Glomerular Filtration Rate (eGFR), and concentrations of glycated hemoglobin (HbA1c), triglycerides, and potassium. The model is independent of day of sampling and type of surgery. Mean MHI was significantly different between patient subgroups with increasing number of comorbidities.CONCLUSION: The MHI reflects severity of comorbidity, enabling objective assessment of a bariatric patient's metabolic health state, regardless day of sampling and surgery type. Next to weight-focused outcome measures like %TWL, the MHI can serve as outcome measure for metabolic health.

Deep Reinforcement Learning (RL) has achieved high success in solving routing problems. However, state-of-the-art deep RL approaches require a considerable amount of data before they reach reasonable performance. This may be acceptable for small problems, but as instances grow bigger, this fact severely limits the applicability of these methods to many real-world instances. In this work, we study a setting where the agent can access data from previously handcrafted heuristics for the Traveling Salesman Problem. In our setting, the agent has access to demonstrations from 2-opt improvement policies. Our goal is to learn policies that can surpass the quality of the demonstrations while requiring fewer samples than pure RL. In this study, we propose to first learn policies with Imitation Learning (IL), leveraging a small set of demonstration data to accelerate policy learning. Afterward, we combine on policy and value approximation updates to improve performance over the expert's performance. We show that our method learns good policies in a shorter time and using less data than classical policy gradient, which does not incorporate demonstration data into RL. Moreover, in terms of solution quality, it performs similarly to other state-of-the-art deep RL approaches.

Background: Tremor severity assessment is an important step for the diagnosis and treatment decision-making of essential tremor (ET) patients. Traditionally, tremor severity is assessed by using questionnaires (e.g. ETRS and QUEST surveys). In this work we assume the possibility of assessing tremor severity using sensor data and computerized analyses. The goal of this work is to assess severity of tremor objectively, to be better able to asses improvement in ET patients due to deep brain stimulation or other treatments. Methods: We collect tremor data by strapping smartphones to the wrists of ET patients. The resulting raw sensor data is then pre-processed to remove any artifact due to patient’s intentional movement. Finally, this data is exploited to automatically build a transparent, interpretable, and succinct fuzzy model for the severity assessment of ET. For this purpose, we exploit pyFUME, a tool for the data-driven estimation of fuzzy models. It leverages the FST-PSO swarm intelligence meta-heuristic to identify optimal clusters in data, reducing the possibility of a premature convergence in local minima which would result in a sub-optimal model. pyFUME was also combined with GRABS, a novel methodology for the automatic simplification of fuzzy rules. Results: Our model is able to assess tremor severity of patients suffering from Essential Tremor, notably without the need for subjective questionnaires nor interviews. The fuzzy model improves the mean absolute error (MAE) metric by 78–81% compared to linear models and by 71–74% compared to a model based on decision trees. Conclusion: This study confirms that tremor data gathered using the smartphones is useful for the constructing of machine learning models that can be used to support the diagnosis and monitoring of patients who suffer from Essential Tremor. The model produced by our methodology is easy to inspect and, notably, characterized by a lower error with respect to approaches based on linear models or decision trees., SCOPUS: ar.j, info:eu-repo/semantics/published

The extremely high turnover of online advertising makes it one of the most important sources of income for many online ad publishers. Advertising through world wide web is mainly performed by Real Time Bidding in which the advertisers and the publishers participate to online auctions for trading the ad slots. Publishers usually set the reserve prices for their ad slots and any winning buyer in the auctions performed by ad exchanges has to pay at least the value of reserve price. Header bidding is a way of real time bidding and it becomes very popular, but how to use it together with advertising Exchanges (AdX) to achieve good revenue for online publishers is not well studied. In this paper, we propose a method that makes use of the historical auction data from header bidding and AdX to learn and optimize the reserve price for AdX. We propose a method based on supervised learning and survival analysis to increase the reserve price. The method assumes no information about current auctions and the bids of header bidding and AdX response are predicted and used to determine the highest possible reserve price. The experiments with real-world auction data show the promising results of our method in increasing the expected revenue of online publishers.

Background The interpretability of results predicted by the machine learning models is vital, especially in the critical fields like healthcare. With the increasingly adoption of electronic healthcare records (EHR) by the medical organizations in the last decade, which accumulated abundant electronic patient data, neural networks or deep learning techniques are gradually being applied to clinical tasks by utilizing the huge potential of EHR data. However, typical deep learning models are black-boxes, which are not transparent and the prediction outcomes of which are difficult to interpret. Methods To remedy this limitation, we propose an attention neural network model for interpretable clinical prediction. In detail, the proposed model employs an attention mechanism to capture critical/essential features with their attention signals on the prediction results, such that the predictions generated by the neural network model can be interpretable. Results We evaluate our proposed model on a real-world clinical dataset consisting of 736 samples to predict readmissions for heart failure patients. The performance of the proposed model achieved 66.7 and 69.1% in terms of accuracy and AUC, respectively, and outperformed the baseline models. Besides, we displayed patient-specific attention weights, which can not only help clinicians understand the prediction outcomes, but also assist them to select individualized treatment strategies or intervention plans. Conclusions The experimental results demonstrate that the proposed model can improve both the prediction performance and interpretability by equipping the model with an attention mechanism.

This paper proposes a Deep Reinforcement Learning (DRL) approach for solving knapsack problem. The proposed method consists of a state aggregation step based on tabular reinforcement learning to extract features and construct states. The state aggregation policy is applied to each problem instance of the knapsack problem, which is used with Advantage Actor Critic (A2C) algorithm to train a policy through which the items are sequentially selected at each time step. The method is a constructive solution approach and the process of selecting items is repeated until the final solution is obtained. The experiments show that our approach provides close to optimal solutions for all tested instances, outperforms the greedy algorithm, and is able to handle larger instances and more flexible than an existing DRL approach. In addition, the results demonstrate that the proposed model with the state aggregation strategy not only gives better solutions but also learns in less timesteps, than the one without state aggregation.

In recent years, it has been demonstrated that checklists can improve patient safety significantly. To facilitate the effective use of checklists in daily practice, both the medical community and the informatics community propose to implement checklists in dynamic checklist applications that can be integrated into the clinical workflow and that is specific to the patient context. However, it is difficult to develop such applications because they are tightly intertwined with the content of specific checklists. We propose a platform that enables access to dynamic checklist applications by configuring the infrastructures provided in the platform. Then, the applications can be developed without time-consuming programming work. We define a number of design criteria regarding point of care and clinical processes by analyzing the existing checklist applications and the lessons learned from implementations. Then, by applying rule-based clinical decision support and workflow management technologies, we design technical mechanisms to satisfy the design criteria. A dynamic checklist application platform is designed based on these mechanisms. Finally, we build a platform in various design cycle iterations, driven by multiple clinical cases. By applying the platform, we develop nine comprehensive dynamic checklist applications with 242 dynamic checklists. The results demonstrate both the feasibility and the overall generic nature of the proposed approach. We propose a novel platform for configuring dynamic checklist applications. This platform satisfies the general requirements and can be easily configured to satisfy different scenarios in which safety checklists are used.

Background Methylmalonic acid (MMA) can detect functional vitamin B12 deficiencies as it accumulates early when intracellular deficits arise. However, impaired clearance of MMA from blood due to decreased glomerular filtration rate (eGFR) also results in elevated plasma MMA concentrations. Alternative to clinical trials, a data mining approach was chosen to quantify and compensate for the effect of decreased eGFR on MMA concentration. Methods Comprehensive data on patient’s vitamin B12, eGFR and MMA concentrations were collected ( n = 2906). The relationship between vitamin B12, renal function (eGFR) and MMA was modelled using weighted multiple linear regression. The obtained model was used to estimate the influence of decreased eGFR on MMA. Clinical impact was examined by comparing the number of patients labelled vitamin B12 deficient with and without adjustment in MMA. Results Adjusting measured MMA concentrations for eGFR in the group of patients with low-normal vitamin B12 concentrations (90–300 pmol/L) showed that the use of unadjusted MMA concentrations overestimates vitamin B12 deficiency by 40%. Conclusions Through a data mining approach, the influence of eGFR on the relation between MMA and vitamin B12 can be quantified and used to correct the measured MMA concentration for decreased eGFR. Especially in the elderly, eGFR-based correction of MMA may prevent over-diagnosis of vitamin B12 deficiency and corresponding treatment.

In Prognostics and Health Management (PHM) sufficient prior observed degradation data is usually critical for Remaining Useful Lifetime (RUL) prediction. Most previous data-driven prediction methods assume that training (source) and testing (target) condition monitoring data have similar distributions. However, due to different operating conditions, fault modes, noise and equipment updates distribution shift exists across different data domains. This shift reduces the performance of predictive models previously built to specific conditions when no observed run-to-failure data is available for retraining. To address this issue, this paper proposes a new data-driven approach for domain adaptation in prognostics using Long Short-Term Neural Networks (LSTM). We use a time window approach to extract temporal information from time-series data in a source domain with observed RUL values and a target domain containing only sensor information. We propose a Domain Adversarial Neural Network (DANN) approach to learn domain-invariant features that can be used to predict the RUL in the target domain. The experimental results show that the proposed method can provide more reliable RUL predictions under datasets with different operating conditions and fault modes. These results suggest that the proposed method offers a promising approach to performing domain adaptation in practical PHM applications.

Many researchers have used fuzzy set theory and fuzzy logic in a variety of applications related to computer science and engineering, given the capability of fuzzy inference systems to deal with uncertainty, represent vague concepts, and connect human language to numerical data. In this work we propose Simpful, a general-purpose and user-friendly Python library designed to facilitate the definition, analysis, and interpretation of fuzzy inference systems. Simpful provides a lightweight Application Programming Interface that allows to intuitively define fuzzy sets and fuzzy rules, and to perform fuzzy inference. Worthy of note, in Simpful the fuzzy rules are specified by means of strings of text written in natural language. We provide here some practical examples to show that Simpful represents a valuable addition to the open-source software that supports fuzzy reasoning.KeywordsDecision support, Fuzzy logic, Fuzzy networks, Modeling and control, Open- source software, Python library

Living in the era of "data deluge" demands for an increase in the application and development of machine learning methods, both in basic and applied research. Among these methods, in the last decades fuzzy inference systems carved out their own niche as (light) grey box models, which are considered more interpretable and transparent than other commonly employed methods, such as artificial neural networks. Although commercially distributed alternatives are available, software able to assist practitioners and researchers in each step of the estimation of a fuzzy model from data are still limited in scope and applicability. This is especially true when looking at software developed in Python, a programming language that quickly gained popularity among data scientists and it is often considered their language of choice. To fill this gap, we introduce pyFUME, a Python library for automatically estimating fuzzy models from data. pyFUME contains a set of classes and methods to estimate the antecedent sets and the consequent parameters of a Takagi-Sugeno fuzzy model from data, and then create an executable fuzzy model exploiting the Simpful library. pyFUME can be beneficial to practitioners, thanks to its pre-implemented and user-friendly pipelines, but also to researchers that want to fine-tune each step of the estimation process.

This case study evaluates to what extent personal preferences can be automatically derived from user event data in an mHealth setting. Based on a theoretical framework, user preferences are described using six classes. Based on this framework, a structure of six Takagi-Sugeno fuzzy inference systems was constructed and evaluated against baseline data from an official survey for measuring the framework's constructs. From this analysis, it was found that user preferences may be derived from user event data using fuzzy modeling with accuracy scores that are higher than a random predictor would typically achieve.

We study a variant of the display-ad allocation problem where an online publisher needs to decide which subset of advertisement slots should be used in order to fulfill guaranteed contracts and which subset should be sold on supply side platforms (SSPs) in order to maximize the expected revenue. Our modeling approach also takes the uncertainty associated with the sale of an impression by an SSP into account. The way that information is revealed over time allows us to model the display-ad allocation problem as a two-stage stochastic program. We refer to our model as the Stochastic Programming (SP) model. Numerical experiments show that the SP model performs well in most cases. We compare the solutions of the SP model with the solutions of an allocation policy (the Priority Assignment (PA) heuristic) that is used in practice. We find that the performance gap between the PA heuristic and the SP model depends on the fraction of total impressions that need to be allocated to the guaranteed contracts. The results suggest that the benefit of using the SP model is highest in periods where the website traffic is high compared with the targets for the guaranteed contracts.

In this work we study a dynamic pricing problem with demand censoring and limited price changes. In our problem there is a seller of a single product that aims to maximize revenue over a finite sales horizon. The seller does not know the form of the mean demand function but does have some limited knowledge. We assume that the seller has a hypothesis set of mean demand functions and that the true mean demand function is an element of this set. Furthermore, the seller faces a business constraint on the number of price changes that is allowed during the sales horizon. More specifically, the number of price changes that the seller is allowed to make is bounded above by a finite integer. We furthermore assume that the seller can only observe the sales (minimum between realized demand and available inventory) and thus that demand is censored. In each period the seller can replenish his inventory to a particular level. The objective of the seller is to set the best price and inventory level in each period of the sales horizon in order to maximize his profit. The profit is determined by the revenue of the sales minus holding costs and costs for lost sales (unsatisfied demand). In determining the best price and inventory level the seller faces and exploration-exploitation trade-off. The seller has to experiment with different prices and inventory levels in order to learn from historical sales data which contains information about market responses to offered prices. On the other hand, the seller also needs to exploit what it has learned and set prices and inventory levels that are optimal given the information collected so far. We propose a heuristic policy for this problem and study its performance using numerical experiments. The results are promising and indicate that the growth rate of regret of the policy is sub-linear with respect to the sales horizon.

Despite the many mHealth solutions available, it remains unclear what their success factors are. Specifically, there has been controversy on the effectiveness of extrinsic rewards. This study evaluates two design elements of an mHealth solution – i.e., social proof and tangible rewards – and their impact on user engagement. During a four-week campaign, a sample of 143 university staff members engaged in a health promotion campaign. Participants were randomly distributed over one of three treatment groups. It was found that the introduction of a sufficiently meaningful, unexpected, and customized extrinsic reward can engage participants significantly more in a health promotion context.