An Improved BPNN Algorithm Based on Deep Learning Technology to analyze the Market Risks of A+H Shares

The backpropagation neural network (BPNN) algorithm of artificial intelligence (AI) is utilized to predict A+H shares price for helping investors reduce the risk of stock investment. First, the genetic algorithm (GA) is used to optimize BPNN, and a model that can predict multi-day stock prices is established. Then, the Principal Component Analysis (PCA) algorithm is introduced to improve the GA-BP model, aiming to provide a practical approach for analyzing the market risks of the A+H shares. The experimental results show that for A shares, the model has the best prediction effect on the price of Bank of China (BC), and the average prediction errors of opening price, maximum price, minimum price, as well as closing price are 0.0236, 0.0262, 0.0294 and 0.0339, respectively. For H shares, the model constructed has the best effect on the price prediction of China Merchants Bank (CMB). The average prediction errors of opening price, maximum price, minimum price and closing price are 0.0276, 0.0422, 0.0194 and 0.0619, respectively.

Pramipexole Enhances Reward Learning by Preserving Value Estimates

Background: Dopamine D2-like receptor agonists show promise as treatments for depression. They are thought to act by altering how individuals learn from rewarding experiences. However, the nature of these reward learning alterations, and the mechanisms by which they are produced is not clear. Reinforcement learning accounts describe three distinct processes that may produce similar changes in reward learning behaviour; increased reward sensitivity, increased inverse decision temperature and decreased value decay. As these processes produce equivalent effects on behaviour, arbitrating between them requires measurement of how expectations and prediction errors are altered. In the present study, we characterised the behavioural effects of a sustained 2-week course of the D2/3/4 receptor agonist pramipexole on reward learning and used fMRI measures of expectation and prediction error to assess which of these three mechanistic processes were responsible for the behavioural effects. Methods: 40 healthy volunteers (Age: 18-43, 50% female) were randomly allocated to receive either two weeks of pramipexole (titrated to 1mg/day) or placebo in a double-blind, between subject design. Participants completed a probabilistic instrumental learning task, in which stimuli were associated with either rewards or losses, before the pharmacological intervention and twice between days 12-15 of the intervention (once with and once without fMRI). Both asymptotic choice accuracy, and a reinforcement learning model, were used to assess reward learning. Results: Behaviourally, pramipexole specifically increased choice accuracy in the reward condition, with no effect in the loss condition. Pramipexole increased the BOLD response in the orbital frontal cortex during the expectation of win trials but decreased the BOLD response to reward prediction errors in the ventromedial prefrontal cortex. This pattern of results indicates that pramipexole enhances choice accuracy by reducing the decay of estimated values during reward learning. Conclusions: The D2-like receptor agonist pramipexole enhances reward learning by preserving learned values. This is a plausible candidate mechanism for pramipexoles observed anti-depressant effect.

Modeling the Metabolic Reductions of a Passive Back-Support Exoskeleton

Despite several attempts to quantify the metabolic savings resulting from the use of passive back-support exoskeletons (BSEs), no study has modeled the metabolic change while wearing an exoskeleton during lifting. The objectives of this study were to: 1) quantify the metabolic reductions due to the VT-Lowe's exoskeleton during lifting; and 2) provide a comprehensive model to estimate the metabolic reductions from using a passive BSE. In this study, 15 healthy adults (13M, 2F) of ages 20 to 34 years (mean=25.33, SD=4.43) performed repeated freestyle lifting and lowering of an empty box and a box with 20% of their bodyweight. Oxygen consumption and metabolic expenditure data were collected. A model for metabolic expenditure was developed and fitted with the experimental data of two prior studies and the without-exoskeleton experimental results. The metabolic cost model was then modified to reflect the effect of the exoskeleton. The experimental results revealed that VT-Lowe's exoskeleton significantly lowered the oxygen consumption by ~9% for an empty box and 8% for a 20% bodyweight box, which corresponds to a net metabolic cost reduction of ~12% and ~9%, respectively. The mean metabolic difference (i.e., without-exo minus with-exo) and the 95% confidence interval were 0.36 and (0.2-0.52) [Watts/kg] for 0% bodyweight, and 0.43 and (0.18-0.69) [Watts/kg] for 20% bodyweight. Our modeling predictions for with-exoskeleton conditions were precise, with absolute freestyle prediction errors of <2.1%. The model developed in this study can be modified based on different study designs, and can assist researchers in enhancing designs of future lifting exoskeletons.

"Learning in reverse: Dopamine errors drive excitatory and inhibitory components of backward conditioning in an outcome-specific manner".

For over two decades, midbrain dopamine was considered synonymous with the prediction error in temporal-difference reinforcement learning. Central to this proposal is the notion that reward-predictive stimuli become endowed with the scalar value of predicted rewards. When these cues are subsequently encountered, their predictive value is compared to the value of the actual reward received allowing for the calculation of prediction errors. Phasic firing of dopamine neurons was proposed to reflect this computation, facilitating the backpropagation of value from the predicted reward to the reward-predictive stimulus, thus reducing future prediction errors. There are two critical assumptions of this proposal: 1) that dopamine errors can only facilitate learning about scalar value and not more complex features of predicted rewards, and 2) that the dopamine signal can only be involved in anticipatory learning in which cues or actions precede rewards. Recent work has challenged the first assumption, demonstrating that phasic dopamine signals across species are involved in learning about more complex features of the predicted outcomes, in a manner that transcends this value computation. Here, we tested the validity of the second assumption. Specifically, we examined whether phasic midbrain dopamine activity would be necessary for backward conditioning- when a neutral cue reliably follows a rewarding outcome. Using a specific Pavlovian-to-Instrumental Transfer (PIT) procedure, we show rats learn both excitatory and inhibitory components of a backward association, and that this association entails knowledge of the specific identity of the reward and cue. We demonstrate that brief optogenetic inhibition of ventral tegmental area dopamine (VTA DA) neurons timed to the transition between the reward and cue, reduces both of these components of backward conditioning. These findings suggest VTA DA neurons are capable of facilitating associations between contiguously occurring events, regardless of the content of those events. We conclude that these data are in line with suggestions that the VTA DA error acts as a universal teaching signal. This may provide insight into why dopamine function has been implicated in a myriad of psychological disorders that are characterized by very distinct reinforcement-learning deficits.

Uncertainty-guided learning with scaled prediction errors in the basal ganglia

To accurately predict rewards associated with states or actions, the variability of observations has to be taken into account. In particular, when the observations are noisy, the individual rewards should have less influence on tracking of average reward, and the estimate of the mean reward should be updated to a smaller extent after each observation. However, it is not known how the magnitude of the observation noise might be tracked and used to control prediction updates in the brain reward system. Here, we introduce a new model that uses simple, tractable learning rules that track the mean and standard deviation of reward, and leverages prediction errors scaled by uncertainty as the central feedback signal. We provide a normative analysis, comparing the performance of the new model with that of conventional models in a value tracking task. We find that the new model has an advantage over conventional models when tested across various levels of observation noise. Further, we propose a possible biological implementation of the model in the basal ganglia circuit. The scaled prediction error feedback signal is consistent with experimental findings concerning dopamine prediction error scaling relative to reward magnitude, and the update rules are found to be consistent with many features of striatal plasticity. Our results span across the levels of implementation, algorithm, and computation, and might have important implications for understanding the dopaminergic prediction error signal and its relation to adaptive and effective learning.

Performance Prediction and Geometry Optimization for Application of Pump as Turbine: A Review

Pump as Turbine (PAT) is a technically and economically effective technology to utilize small/mini/micro/pico hydropower, especially in rural areas. There are two main subjects that influence the selection and application of PAT. On the one hand, manufacturers of pumps will not provide their characteristics under the turbine mode, which requires performance prediction methods. On the other hand, PAT efficiency is always slightly lower than that of pump, which requires further geometry optimization. This literature review summarized published research studies related to performance prediction and geometry optimization, aimed at guiding for selection and optimization of PAT. Currently, there exist four categories of performance prediction methods, namely, using BEP (Best Efficiency Point), using specific speed, loss modeling, and polynomial fitting. The using BEP and loss modeling methods are based on theoretical analysis, while using specific speed and polynomial fitting methods require statistical fitting. The prediction errors of published methods are within ±10% mostly. For geometry optimization, investigations mainly focus on impeller diameter and blade geometry. The influence of impeller trimming, blade rounding, blade wrap angle, blade profile, blade number, blade trailing edge position, and guide vane number has been studied. Among published methods, the blade rounding and forward-curved impellers are the most effective and feasible techniques.

Lifetime and Aging Degradation Prognostics for Lithium-ion Battery Packs Based on a Cell to Pack Method

Aging diagnosis of batteries is essential to ensure that the energy storage systems operate within a safe region. This paper proposes a novel cell to pack health and lifetime prognostics method based on the combination of transferred deep learning and Gaussian process regression. General health indicators are extracted from the partial discharge process. The sequential degradation model of the health indicator is developed based on a deep learning framework and is migrated for the battery pack degradation prediction. The future degraded capacities of both battery pack and each battery cell are probabilistically predicted to provide a comprehensive lifetime prognostic. Besides, only a few separate battery cells in the source domain and early data of battery packs in the target domain are needed for model construction. Experimental results show that the lifetime prediction errors are less than 25 cycles for the battery pack, even with only 50 cycles for model fine-tuning, which can save about 90% time for the aging experiment. Thus, it largely reduces the time and labor for battery pack investigation. The predicted capacity trends of the battery cells connected in the battery pack accurately reflect the actual degradation of each battery cell, which can reveal the weakest cell for maintenance in advance.

An airfoil geometric-feature extraction and discrepant data fusion learning method

The perception of geometric-features of airfoils is the basis in aerodynamic area for performance prediction, parameterization, aircraft inverse design, etc. There are three approaches to percept the geometric shape of airfoils, namely manual design of airfoil geometry parameters, polynomial definition and deep learning. The first two methods directly define geometric-features or polynomials of airfoil curves, but the number of extracted features is limited. Deep learning algorithms can extract a large number of potential features (called latent features). However, the features extracted by deep learning lack explicit geometrical meaning. Motivated by the advantages of polynomial definition and deep learning, we propose a geometric-feature extraction method (named Bézier-based feature extraction, BFE) for airfoils, which consists of two parts: manifold metric feature extraction and geometric-feature fusion encoder (GF encoder). Manifold metric feature extraction, with the help of the Bézier curve, captures manifold metrics (a sort of geometric-features) from tangent space of airfoil curves, and the GF-encoder combines airfoil coordinate data and manifold metrics together to form novel fused geometric-features. To validate the feasibility of the fused geometric-features, two experiments based on the public UIUC airfoil dataset are conducted. Experiment I is used to extract manifold metrics of airfoils and export the fused geometric-features. Experiment II, based on the Multi-task learning (MTL), is used to fuse the discrepant data (i.e., the fused geometric-features and the flight conditions) to predict the aerodynamic performance of airfoils. The results show that the BFE can generate more smooth and realistic airfoils than Auto-Encoder, and the fused geometric-features extracted by BFE can be used to reduce the prediction errors of C L and C D .

Study on Motion Management of Lung Cancer Treated by CyberKnife: Based on Tumor Position

Purpose: We investigated the movement characteristics of lung cancers and the clinical accuracy of tracking lung tumors with Synchrony Respiratory Tracking System (SRTs) during the CyberKnife treatment. We also explored the influencing factors of accuracy. These data provided the appropriate expansion margins of patients with different respiratory characteristics, which was helpful to realize the personalized design of treatment plans of CyberKnife. Methods and Materials: 73 patients with lung cancer treated with CyberKnife SRTs were selected retrospectively for this study. The patient's age, gender, respiratory characteristics and tumor datas (tumor size, anatomical position and geometric position) were recorded. During treatment, the deviation was checked every 45 s and compensated by the synchronous respiratory tracking system.Results: The total mean motion amplitudes and standard deviations of lung tumors in superior-inferior (SI), left-right (LR), and anterior-posterior (AP) directions were 4.15 ± 3.47 mm, 3.98 ± 3.21 mm and 3.79 ± 2.73 mm, respectively. The overall mean correlation errors and standard deviations were 0.86 ± 0.45 mm, 1.04 ± 0.76 mm and 0.70 ± 0.47 mm, respectively. The overall mean prediction errors and standard deviations were 0.18 ± 0.17 mm, 0.35 ± 0.39 mm and 0.35 ± 0.42 mm, respectively. The correlation errors of LR direction were less correlated with the geometric position of the tumor (r = 0.38), and not correlated with the anatomical position of the tumor (r < 0.3). The prediction errors were moderately correlated with the respiratory amplitude (r = 0.588), and less correlated with the baseline drift and the motion amplitude of the tumor (r = 0.407 and 0.365, respectively).Conclusions: The patient’s respiratory amplitude, the tumor motion amplitude, the tumor baseline drift and geometric position were the main factors affecting the tracking accuracy. Tumors at different geometric positions should be treated differently to ensure sufficient dose coverage of the lung tumor target.