Overestimation Bias Research Articles

A proxy measure of striatal dopamine predicts individual differences in temporal precision

The perception of time is characterized by pronounced variability across individuals, with implications for a diverse array of psychological functions. The neurocognitive sources of this variability are poorly understood, but accumulating evidence suggests a role for inter-individual differences in striatal dopamine levels. Here we present a pre-registered study that tested the predictions that spontaneous eyeblink rates, which provide a proxy measure of striatal dopamine availability, would be associated with aberrant interval timing (lower temporal precision or overestimation bias). Neurotypical adults (N = 69) underwent resting state eye tracking and completed visual psychophysical interval timing and control tasks. Elevated spontaneous eyeblink rates were associated with poorer temporal precision but not with inter-individual differences in perceived duration or performance on the control task. These results signify a role for striatal dopamine in variability in human time perception and can help explain deficient temporal precision in psychiatric populations characterized by elevated dopamine levels.

Protein requirements and provision in hospitalised COVID-19 ward and ICU patients: Agreement between calculations based on body weight and height, and measured bioimpedance lean body mass.

BackgroundA large proportion of hospitalised COVID-19 patients are overweight. There is no consensus in the literature on how lean body mass (LBM) can best be estimated to adequately guide nutritional protein recommendations in hospitalised patients who are not at an ideal weight. We aim to explore which method best agrees with lean body mass as measured by bioelectric impedance (LBMBIA) in this population.MethodsLBM was calculated by five commonly used methods for 150 hospitalised COVID-19 patients previously included in the BIAC-19 study; total body weight, regression to a BMI of 22.5, regression to BMI 27.5 when BMI>30, and the equations described by Gallagher and the ESPEN ICU guideline. Error–standard plots were used to assess agreement and bias compared to LBMBIA. The actual protein provided to ICU patients during their stay was compared to targets set using LBMBIA and LBM calculated by other methods.ResultsAll methods to calculate LBM suffered from overestimation, underestimation, fixed- and proportional bias and wide limits of agreement compared to LBMBIA. Bias was inconsistent across sex and BMI subgroups. Twenty-eight ICU patients received a mean of 51.19 (95%-BCa CI 37.1;64.1) grams of protein daily, accumulating to a mean of 61.6% (95%-BCa CI 43.2;80.8) of TargetBIA during their ICU stay. The percentage received of the target as calculated by the LBMGallagher method for males was the only one to not differ significantly from the percentage received of TargetBIA (mean difference 1.4% (95%-BCa CI -1.3;4.6) p = 1.0).ConclusionsWe could not identify a mathematical method for calculating LBM that had an acceptable agreement with LBM as derived from BIA for males and females across all BMI subgroups in our hospitalised COVID-19 population. Consequently, discrepancies when assessing the adequacy of protein provision in ICU patients were found. We strongly advise using baseline LBMBIA to guide protein dosing if possible. In the absence of BIA, using a method that overestimates LBM in all categories may be the only way to minimise underdosing of nutritional protein.Trial registrationThe protocol of the BIAC-19 study, of which this is a post-hoc sub-analysis, is registered in the Netherlands Trial Register (number NL8562).

The Accuracy of Precipitation Forecasts at Timescales of 1–15 Days in the Volta River Basin

Medium-range (1–15 day) precipitation forecasts are increasingly available from global weather models. This study presents evaluation of the Global Forecast System (GFS) for the Volta river basin in West Africa. The evaluation was performed using two satellite-gauge merged products: NASA’s Integrated Multi-satellitE Retrievals (IMERG) “Final Run” satellite-gauge merged rainfall observations, and the University of California Santa Barbara’s Climate Hazard’s group Infrared Precipitation with Stations (CHIRPS). The performance of GFS depends on the climate zone, with underestimation bias in the dry Sahel climate, overestimation bias in the wet Guinea Coastal climate, and relatively no bias in the moderately wet Savannah climate. Averaging rainfall over the watershed of the Akosombo dam (i.e., averaging across all three climate zones), the GFS forecast indicates low skill (Kling-Gupta Efficiency KGE = 0.42 to 0.48) for the daily, 1-day, lead GFS forecast, which deteriorates further as the lead time increases. A sharp decrease in KGE occurred between 6 to 10 days. Aggregating the forecasts over long timescales improves the accuracy of the GFS forecasts. On a 15-day accumulation timescale, GFS shows higher skills (KGE = 0.74 to 0.88).

Deterministic policy optimization with clipped value expansion and long-horizon planning

Model-based reinforcement learning (MBRL) approaches have demonstrated great potential in handling complex tasks with high sample efficiency. However, MBRL struggles with asymptotic performance compared to model-free reinforcement learning (MFRL). In this paper, we present a long-horizon policy optimization method, namely model-based deterministic policy gradient (MBDPG), for efficient exploitation of the learned dynamics model through multi-step gradient information. First, we approximate the dynamics of the environment with a parameterized linear combination of an ensemble of Gaussian distributions. Moreover, the dynamics model is equipped with a memory module and trained on a multi-step prediction task to reduce cumulative error. Second, successful experience is used to guide the policy at the early stage of training to avoid ineffective exploration. Third, a clipped double value network is expanded in the learned dynamics to reduce overestimation bias. Finally, we present a deterministic policy gradient approach in the model that backpropagates multi-step gradient along the imagined trajectories. Our method shows higher sampling efficiency than the state-of-the-art MFRL methods while maintaining better convergence performance and time efficiency compared to the SOAT MBRL.

COVID-19 Vaccination Willingness and Acceptability in Multiple Sclerosis Patients: A Cross Sectional Study in Iran.

Multiple sclerosis (MS) is a chronic, predominantly immune-mediated degenerative disease of the central nervous system. Due to prolonged use of immunomodulatory and immunosuppressive medications, vaccine hesitancy could be common among MS patients. Our main aim in the current study was to evaluate the willingness and acceptability of COVID-19 vaccination in patients with MS. In our multicenter cross-sectional questionnaire-based clinical study, 892 patients completed the questionnaire between May to June 2021. The questionnaire consisted of demographic data, MS disease-related factors, history of COVID-19 infection/vaccination, and any existing comorbidities. Statistical analysis was performed using SPSS software version 19. Overall, 68% of the participants expressed willingness to be vaccinated. Major causes of vaccine refusal in our patients were the fear of reducing the efficacy of disease modifying drugs (DMDs) upon vaccination as well as distrusting the vaccines and overestimation bias in the power of their innate immunity and potential COVID-19 resistance. Some demographic factors affected vaccination enthusiasm in our study. Our findings did not show significant correlation between the age and comorbidity and vaccine willingness. Only one-third of our patients received their vaccine information from healthcare providers. The majority of them received these data from official broadcasting channels and social media. However, despite several concerns, the willingness of COVD-19 vaccination in the Iranian MS patients is remarkable.

Performance of the Global Forecast System's medium-range precipitation forecasts in the Niger river basin using multiple satellite-based products

Abstract. Accurate weather forecast information has the potential to improve water resources management, energy, and agriculture. This study evaluates the accuracy of medium-range (1–15 d) precipitation forecasts from the Global Forecast System (GFS) over watersheds of eight major dams (Selingue Dam, Markala Dam, Goronyo Dam, Bakolori Dam, Kainji Dam, Jebba Dam, Dadin Kowa Dam, and Lagdo Dam) in the Niger river basin using NASA's Integrated Multi-satellitE Retrievals (IMERG) Final Run merged satellite gauge rainfall observations. The results indicate that the accuracy of GFS forecast varies depending on climatic regime, lead time, accumulation timescale, and spatial scale. The GFS forecast has large overestimation bias in the Guinea region of the basin (wet climatic regime), moderate overestimation bias in the Savannah region (moderately wet climatic regime), but has no bias in the Sahel region (dry climate). Averaging the forecasts at coarser spatial scales leads to increased forecast accuracy. For daily rainfall forecasts, the performance of GFS is very low for almost all watersheds, except for Markala and Kainji dams, both of which have much larger watershed areas compared to the other watersheds. Averaging the forecasts at longer timescales also leads to increased forecast accuracy. The GFS forecasts, at 15 d accumulation timescale, have better performance but tend to overestimate high rain rates. Additionally, the performance assessment of two other satellite products was conducted using IMERG Final estimates as reference. The Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS) merged satellite gauge product has similar rainfall characteristics to IMERG Final, indicating the robustness of IMERG Final. The IMERG Early Run satellite-only rainfall product is biased in the dry Sahel region; however, in the wet Guinea and Savannah regions, IMERG Early Run outperforms GFS in terms of bias.

Evaluation of Global Forecast System (GFS) Medium-Range Precipitation Forecasts in the Nile River Basin

Abstract Reliable weather forecasts are valuable in a number of applications, such as agriculture, hydropower, and weather-related disease outbreaks. Global weather forecasts are widely used, but detailed evaluation over specific regions is paramount for users and operational centers to enhance the usability of forecasts and improve their accuracy. This study presents evaluation of the Global Forecast System (GFS) medium-range (1–15 day) precipitation forecasts in the nine subbasins of the Nile basin using NASA’s Integrated Multisatellite Retrievals (IMERG) Final Run satellite–gauge merged rainfall observations. The GFS products are available at a temporal resolution of 3–6 h and a spatial resolution of 0.25°, and the version-15 products are available since 12 June 2019. GFS forecasts are evaluated at a temporal scale of 1–15 days, a spatial scale from 0.25° to all the way to the subbasin scale, and for a period of one year (15 June 2019–15 June 2020). The results show that performance of the 1-day lead daily basin-averaged GFS forecast performance, as measured through the modified Kling–Gupta efficiency (KGE), is poor (0 < KGE < 0.5) for most of the subbasins. The factors contributing to the low performance are 1) large overestimation bias in watersheds located in wet climate regimes in the northern hemispheres (Millennium watershed, Upper Atbara and Setit watershed, and Khashm El Gibra watershed), and 2) lower ability in capturing the temporal dynamics of watershed-averaged rainfall that have smaller watershed areas (Roseires at 14 110 km2 and Sennar at 13 895 km2). GFS has better bias for watersheds located in the dry parts of the Northern Hemisphere or wet parts of the Southern Hemisphere, and better ability in capturing the temporal dynamics of watershed-average rainfall for large watershed areas. IMERG Early has better bias than GFS forecast for the Millennium watershed but still comparable and worse bias for the Upper Atbara and Setit and Khashm El Gibra watersheds. The variation in the performance of the IMERG Early could be partly explained by the number of rain gauges used in the reference IMERG Final product, as 16 rain gauges were used for the Millennium watershed but only one rain gauge over each Upper Atbara and Setit and Khashm El Gibra watershed. A simple climatological bias correction of IMERG Early reduces in the bias in IMERG Early over most watersheds, but not all watersheds. We recommend exploring methods to increase the performance of GFS forecasts, including postprocessing techniques through the use of both near-real-time and research-version satellite rainfall products.

Evaluation and Bias correction of Tropical Applications of Meteorology using SATellite (TAMSAT) daily rainfall estimates over the data-scarce region of Southern Ethiopia

Precipitation data is an intrinsic parameter of rainfall-runoff simulation. since it is strongly hooked into the accuracy of the spatial and temporal representation of the precipitation. In regions where rainfall stations are scarce, additional data sources could also be needed. Satellite platforms have provided a satisfactory alternative because of their global coverage. Tropical Applications of Meteorology using SATellite (TAMSAT) is one such data source provided by the research group of the University of Reading, UK. This dataset provides rainfall estimates across Africa which captures the rainfall pattern in GenaleDawa River basin (southern Ethiopia) but with a magnitude bias of over-estimation and under-estimations. In this study, magnitude bias correction of the TAMSAT dataset with a linear scaling technique resulted in a rainfall grid of the area with ~4 km spatial resolution of a 35 year (1983–2017) daily rainfall dataset. Intercomparison between Satellite rainfall product and observed data were done using point to grid method selecting nine representative metrological stations.The average correlation R is 0.45 and the average NS is 0.028 and the Categorical index POD, FAR, FB and HSS, values were 0.49,0.4,0.84, and 0.4) before bias correction. After bias correction, these values were improved to the average value of R=0.87 and NS =0.764 and POD, FAR, FB, and HSS of (0.71,0.22, 0.92, and 0.66 respectively

Divergent responses of maize yield to precipitation in the United States

How maize yield response to precipitation varies across a large spatial scale is unclear compared with the well-understood temperature response, even though precipitation change is more erratic with greater spatial heterogeneity. This study provides a spatial-explicit quantification of maize yield response to precipitation in the contiguous United States and investigates how precipitation response is altered by natural and human factors using statistical and crop model data. We find the precipitation responses are highly heterogeneous with inverted-U (40.3%) being the leading response type, followed by unresponsive (30.39%), and linear increase (28.6%). The optimal precipitation threshold derived from inverted-U response exhibits considerable spatial variations, which is higher under wetter, hotter, and well-drainage conditions but lower under drier, cooler, and poor-drainage conditions. Irrigation alters precipitation response by making yield either unresponsive to precipitation or having lower optimal thresholds than rainfed conditions. We further find that the observed precipitation responses of maize yield are misrepresented in crop models, with a too high percentage of increase type (59.0% versus 29.6%) and an overestimation in optimal precipitation threshold by ∼90 mm. These two factors explain about 30% and 85% of the inter-model yield overestimation biases under extreme rainfall conditions. Our study highlights the large spatial heterogeneity and the key role of human management in the precipitation responses of maize yield, which need to be better characterized in crop modeling and food security assessment under climate change.

Twin-Delayed Deep Deterministic Policy Gradient for Low-Frequency Oscillation Damping Control

Due to the large scale of power systems, latency uncertainty in communications can cause severe problems in wide-area measurement systems. To resolve this issue, a significant amount of past work focuses on using emerging technology, including machine learning methods such as Q-learning, for addressing latency issues in modern controls. Although the method can deal with the stochastic characteristics of communication latency, the Q-values can be overestimated in Q-learning methods, leading to high bias. To address the overestimation bias issue, we redesign the learning structure of the deep deterministic policy gradient (DDPG). Then we develop a damping control twin-delayed deep deterministic policy gradient method to handle the damping control issue under unknown latency in the power network. The purpose is to address the damping control issue under unknown latency in the power network. This paper will create a novel reward algorithm, taking into account the machine speed deviation, the episode termination prevention, and the feedback from action space. In this way, the system optimally damps down frequency oscillations while maintaining the system’s stability and reliable operation within defined limits. The simulation results verify the proposed algorithm in various perspectives, including the latency sensitivity analysis under high renewable energy penetration and the comparison with conventional and machine learning control algorithms. The proposed method shows a fast learning curve and good control performance under varying communication latency.

What Is the Most Reliable Method of Measuring Glenoid Bone Loss in Anterior Glenohumeral Instability? A Cadaveric Study Comparing Different Measurement Techniques for Glenoid Bone Loss

Background: Preoperative quantification of bone loss has a significant effect on surgical decision making and patient outcomes. Various measurement techniques for calculating glenoid bone loss have been proposed in the literature. To date, no studies have directly compared measurement techniques to determine which technique, if any, is the most reliable. Purpose/Hypothesis: To identify the most consistent and accurate techniques for measuring glenoid bone loss in anterior glenohumeral instability. Our hypothesis was that linear measurement techniques would have lower consistency and accuracy than surface area and statistical shape model–based measurement techniques. Study Design: Controlled laboratory study. Methods: In 6 fresh-frozen human shoulders, 3 incremental bone defects were sequentially created resulting in a total of 18 glenoid bone defect samples. Analysis was conducted using 2D and 3D computed tomography (CT) en face images. A total of 6 observers (3 experienced and 3 with less experience) measured the bone defect of all samples with Horos imaging software using 5 common methods. The methods included 2 linear techniques (Shaha, Griffith), 2 surface techniques (Barchilon, PICO), and 1 statistical shape model formula (Giles). Intraclass correlation (ICC) using a consistency model was used to determine consistency between observers for each of the measurement methods. Paired t tests were used to calculate the accuracy of each measurement technique relative to physical measurement. Results: For the more experienced observers, all methods indicated good consistency (ICC > 0.75; range, 0.75-0.88), except the Shaha method, which indicated moderate consistency (0.65 < ICC < 0.75; range, 0.65-0.74). Estimated consistency among the experienced observers was better for 2D than 3D images, although the differences were not significant (intervals contained 0). For less experienced observers, the Giles method in 2D had the highest estimated consistency (ICC, 0.88; 95% CI, 0.76-0.95), although Giles, Barchilon, Griffith, and PICO methods were not statistically different. Among less experienced observers, the 2D images using Barchilon and Giles methods had significantly higher consistency than the 3D images. Regarding accuracy, most of the methods statistically overestimated the actual physical measurements by a small amount (mean within 5%). The smallest bias was observed for the 2D Barchilon measurements, and the largest differences were observed for Giles and Griffith methods for both observer types. Conclusion: Glenoid bone loss calculation presents variability depending on the measurement technique, with different consistencies and accuracies. We recommend use of the Barchilon method by surgeons who frequently measure glenoid bone loss, because this method presents the best combined consistency and accuracy. However, for surgeons who measure glenoid bone loss occasionally, the most consistent method is the Giles method, although an adjustment for the overestimation bias may be required. Clinical Relevance: The Barchilon method for measuring bone loss has the best combined consistency and accuracy for surgeons who frequently measure bone loss.

Perceptual Accuracy for Sexual Rejection in Romantic Relationships.

In the current research, we investigated perceptual accuracy in romantic partners' detection of sexual rejection. In two daily diary studies of predominantly cisgender heterosexual couples, we examined patterns of accuracy and bias concerning both the degree of sexual disinterest (Study 1; N = 98 couples) and occurrence of sexual rejection (Study 2; N = 115 couples), as well as how these perceptions were associated with satisfaction. Using a multi-method approach to capture both continuous and categorical operationalizations of sexual rejection (Study 1: truth and bias; Study 2: quasi-signal detection), we found that people were both accurate and biased in their perceptions of partner rejection. Across studies, results showed that people demonstrated general tracking accuracy in detecting a partner's sexual rejection, but they also overestimated the degree and occurrence of this rejection. Additionally, this overestimation bias was associated with lower daily relationship and sexual satisfaction. Overall, our findings highlight the importance of dyadic perceptions of sexual rejection in shaping daily relationship and sexual functioning.

Action Candidate Based Clipped Double Q-learning for Discrete and Continuous Action Tasks

Double Q-learning is a popular reinforcement learning algorithm in Markov decision process (MDP) problems. Clipped Double Q-learning, as an effective variant of Double Q-learning, employs the clipped double estimator to approximate the maximum expected action value. Due to the underestimation bias of the clipped double estimator, performance of clipped Double Q-learning may be degraded in some stochastic environments. In this paper, in order to reduce the underestimation bias, we propose an action candidate based clipped double estimator for Double Q-learning. Specifically, we first select a set of elite action candidates with the high action values from one set of estimators. Then, among these candidates, we choose the highest valued action from the other set of estimators. Finally, we use the maximum value in the second set of estimators to clip the action value of the chosen action in the first set of estimators and the clipped value is used for approximating the maximum expected action value. Theoretically, the underestimation bias in our clipped Double Q-learning decays monotonically as the number of the action candidates decreases. Moreover, the number of action candidates controls the trade-off between the overestimation and underestimation biases. In addition, we also extend our clipped Double Q-learning to continuous action tasks via approximating the elite continuous action candidates. We empirically verify that our algorithm can more accurately estimate the maximum expected action value on some toy environments and yield good performance on several benchmark problems.

The Control Method of Twin Delayed Deep Deterministic Policy Gradient with Rebirth Mechanism to Multi-DOF Manipulator

As a research hotspot in the field of artificial intelligence, the application of deep reinforcement learning to the learning of the motion ability of a manipulator can help to improve the learning of the motion ability of a manipulator without a kinematic model. To suppress the overestimation bias of values in Deep Deterministic Policy Gradient (DDPG) networks, the Twin Delayed Deep Deterministic Policy Gradient (TD3) was proposed. This paper further suppresses the overestimation bias of values for multi-degree of freedom (DOF) manipulator learning based on deep reinforcement learning. Twin Delayed Deep Deterministic Policy Gradient with Rebirth Mechanism (RTD3) was proposed. The experimental results show that RTD3 applied to multi degree freedom manipulators is in place, with an improved learning ability by 29.15% on the basis of TD3. In this paper, a step-by-step reward function is proposed specifically for the learning and innovation of the multi degree of freedom manipulator’s motion ability. The view of continuous decision-making and process problem is used to guide the learning of the manipulator, and the learning efficiency is improved by optimizing the playback of experience. In order to measure the point-to-point position motion ability of a manipulator, a new evaluation index based on the characteristics of the continuous decision process problem, energy efficiency distance, is presented in this paper, which can evaluate the learning quality of the manipulator motion ability by a more comprehensive and fair evaluation algorithm.

Re-evaluation of the comparative effectiveness of bootstrap-based optimism correction methods in the development of multivariable clinical prediction models

BackgroundMultivariable prediction models are important statistical tools for providing synthetic diagnosis and prognostic algorithms based on patients’ multiple characteristics. Their apparent measures for predictive accuracy usually have overestimation biases (known as ‘optimism’) relative to the actual performances for external populations. Existing statistical evidence and guidelines suggest that three bootstrap-based bias correction methods are preferable in practice, namely Harrell’s bias correction and the .632 and .632+ estimators. Although Harrell’s method has been widely adopted in clinical studies, simulation-based evidence indicates that the .632+ estimator may perform better than the other two methods. However, these methods’ actual comparative effectiveness is still unclear due to limited numerical evidence.MethodsWe conducted extensive simulation studies to compare the effectiveness of these three bootstrapping methods, particularly using various model building strategies: conventional logistic regression, stepwise variable selections, Firth’s penalized likelihood method, ridge, lasso, and elastic-net regression. We generated the simulation data based on the Global Utilization of Streptokinase and Tissue plasminogen activator for Occluded coronary arteries (GUSTO-I) trial Western dataset and considered how event per variable, event fraction, number of candidate predictors, and the regression coefficients of the predictors impacted the performances. The internal validity of C-statistics was evaluated.ResultsUnder relatively large sample settings (roughly, events per variable ≥ 10), the three bootstrap-based methods were comparable and performed well. However, all three methods had biases under small sample settings, and the directions and sizes of biases were inconsistent. In general, Harrell’s and .632 methods had overestimation biases when event fraction become lager. Besides, .632+ method had a slight underestimation bias when event fraction was very small. Although the bias of the .632+ estimator was relatively small, its root mean squared error (RMSE) was comparable or sometimes larger than those of the other two methods, especially for the regularized estimation methods.ConclusionsIn general, the three bootstrap estimators were comparable, but the .632+ estimator performed relatively well under small sample settings, except when the regularized estimation methods are adopted.

To Plan or Not to Plan? A Mixed-Methods Diary Study Examining When, How and Why Knowledge Work Planning is Inaccurate

Reliable and accurate planning is essential for modern knowledge workers. However, there is limited insight about when, how and why planning is inaccurate, and the circumstances in which those inaccuracies are troublesome. To investigate this, we asked 20 academics to keep a diary for a single work day. They estimated the duration of the tasks they wanted to achieve at the start of the day and noted down in detail the tasks they actually achieved during the day. Semi-structured interviews were conducted to complement this diary data. The diaries showed that some tasks, such as email and coding, were more susceptible to time underestimation bias while other tasks, such as writing and planning, were more susceptible to time overestimation bias in planning. Based on interviews, a typology of common reasons for delays in planned daily work is presented. It suggests that vague and optimistic planning leads to the observed discrepancy between planned and actual work. Finally, interviews suggested that participants adopted four planning strategies that vary in the frequency of planning, from minimal planning to daily, weekly and multi-level planning. We close by discussing ways support systems for accurate planning can be better designed for different use cases.

Why Do Actions in Slow Motion Appear to Last Longer? On the Effect of Video Speed Information.

When displayed in slow motion, actions are often perceived longer compared with original speed. However, it remains to be determined why this bias exists. Is it possible that the bias emerges because participants underestimate the factor by which a video was slowed down and hence arrive at erroneous conclusions about the original duration? If true, providing explicit information about the respective video speed should eliminate this slow motion effect. To scrutinize the nature of this bias, participants rated the original duration of sports actions displayed at original speed or slow motion. Results revealed the expected overestimation bias consisting in longer ratings with increasing slow motion. However, the bias disappeared when information about the current video speed was provided. The observations suggest an influence of knowledge about video playback speed on cognitive-evaluative processes which may hold important implications for future research and practice.

Revisiting the inflation perception conundrum

Households' inflation perceptions and expectations play a key role in many dynamic macroeconomic and monetary models and are important for the ability of central banks to reach their objective of price stability. This paper revisits the issue of overestimation bias in inflation perceptions and expectations based on new and unique microdata from the Danish part of the EU-Harmonised Consumer Expectations Survey linked to rich household-level data from administrative registers. The analysis shows that accounting for even several of the household characteristics and social gradients usually addressed in the literature is far from sufficient to explain the inflation perception bias. Furthermore, we find that respondents participating in the survey more than once tend to be persistent in their degree of perception bias and that overpessimistic households have larger perception bias than other households. This indicates that inflation perception bias is related to fundamental personality traits. Finally, households' expectations of the future inflation level tend to be mean reverting and associated with the same types of bias as inflation perceptions.

Multi-Agent Reinforcement Learning based on Value Distribution

Reinforcement learning in a multi-agent setting is very important for real-world applications, but it brings more challenges than those in a single-agent environment. In the multi-agent setting, the agent generally has a bias of overestimation on the value function. In our work, we pay attention to the issue of overestimation bias with continuous actions in the multi-agent learning environment. We propose a method to reduce this bias by adopting the distributional perspective on reinforcement learning. We combine it within the framework of off-policy learning Actor-Critic and propose a novel approach Multi-Agent Deep Distributional Deterministic Policy Gradient (MAD3PG). We empirically evaluate it in three competitive and cooperative multi-agent settings. Our results show that in a series of difficult motor tasks the agents trained by MAD3PG significantly outperforms existing benchmark.

The validity of the Push Band 2.0 to determine speed and power during progressively loaded squat jumps

ABSTARCT The PUSH band 2.0 is a wearable technology used to measure mean and peak velocity and power in strength-based movements. The agreement between the PUSH band 2.0 and the criterion measure (force plates) during progressively loaded squat jumps was assessed. Fifteen participants performed 3 squat jumps at increasing loads. Linear regression and Bland–Altman plots assessed data simultaneously recorded from both devices. Mean velocity and power showed deviation from the identity line and an overestimation of 7.40% and 25%, respectively. Peak velocity and power showed an overestimation of 14% and underestimation of 6%, respectively. The results support the use of Push Band 2.0 to measure velocity during ballistic squat movements. However, errors in power measurement are greater than acceptable to support in-field use. While peak velocity maintains a consistent overestimation bias across various velocities, mean velocity error increases at higher velocities and can only be considered valid at slow velocities.