Forecasting stock prices plays an important role in setting a trading strategy or determining the appropriate timing for buying or selling a stock. We propose a model, called the feature fusion long ...short-term memory-convolutional neural network (LSTM-CNN) model, that combines features learned from different representations of the same data, namely, stock time series and stock chart images, to predict stock prices. The proposed model is composed of LSTM and a CNN, which are utilized for extracting temporal features and image features. We measure the performance of the proposed model relative to those of single models (CNN and LSTM) using SPDR S&P 500 ETF data. Our feature fusion LSTM-CNN model outperforms the single models in predicting stock prices. In addition, we discover that a candlestick chart is the most appropriate stock chart image to use to forecast stock prices. Thus, this study shows that prediction error can be efficiently reduced by using a combination of temporal and image features from the same data rather than using these features separately.
Convectively generated gravity waves (CGGWs) are important for numerical weather prediction due to their effect on the quasi‐biennial oscillation (QBO) in the stratosphere. Using global ECMWF IFS ...simulations at TCo7999 (or 1.25 km), TCo2559 (or 3.9 km) and TCo1279 (or 7.8 km) horizontal resolutions, sensitivity of resolved CGGWs to the horizontal resolution and to the explicit versus parametrized representation of deep convection is elucidated during the westerly shear phase of the QBO. Parametrized deep convection is found to inhibit CGGWs, resulting in a twofold reduction in CGGW forcing. When deep convection is explicitly resolved, the total CGGW forcing is almost unchanged across the horizontal resolutions. However, the contribution of long and mesoscale CGGWs (with horizontal wavelengths 100 km ≤λh<1,900 km) to the total CGGW forcing decreases and the contribution of smaller‐scale CGGWs (with λh<100 km) increases as the horizontal resolution increases. At the maximum CGGW forcing altitude, at TCo7999 resolution 43% of the total CGGW forcing is due to long and mesoscale waves, whereas at TCo2559 and TCo1279 resolutions their contribution is 70% and 90%, respectively. While CGGW forcing from long and mesoscale waves is similar at TCo7999 resolution with explicit deep convection and at TCo1279 resolution with parametrized deep convection, CGGW forcing from these waves is artificially enhanced at TCo1279 and TCo2559 resolutions with explicit deep convection. This is due to the explicit deep convection being too strong and having too much variance for 100 km ≤λh<1,900 km. Therefore, parametrizations of deep convection and CGGWs (to account for forcing from waves with λh<100 km) are required even at TCo2559 resolution. Additionally, resolved CGGW forcing at TCo7999 resolution is examined for the easterly shear phase of the QBO; similar to the westerly shear phase, the smaller‐scale waves contribute >55% to the total CGGW forcing at the maximum CGGW forcing altitude.
Resolved convectively generated gravity wave (GW) forcing in the tropical stratosphere is quantified at an unprecedented global horizontal resolution of O(1 km). Small‐scale GWs with horizontal wavelengths < 100 km are as important as (or more important than) the long and mesoscale GWs with horizontal wavelengths > 100 km for driving the QBO. Deep convection parametrization inhibits resolved convective GW generation but if deep convection parametrization is switched off at 4–9 km horizontal resolution, power in long and mesoscale GWs is artificially enhanced.
•Proposed a new modularized forecasting framework (ModAugNet) for stock market.•ModAugNet has two LSTM modules: overfitting Prevention Module and Prediction Module.•Found that Prevention Module helps ...to prevent network from overfitting when training.•Verified that ModAugNet significantly outperformed a model without Prevention Module.•Showed that test performance solely depends on test input of Prediction Module.
Forecasting a financial asset's price is important as one can lower the risk of investment decision- making with accurate forecasts. Recently, the deep neural network is popularly applied in this area of research; however, it is prone to overfitting owing to limited availability of data points for training. We propose a novel data augmentation approach for stock market index forecasting through our ModAugNet framework, which consists of two modules: an overfitting prevention LSTM module and a prediction LSTM module.
The performance of the proposed model is evaluated using two different representative stock market data (S&P500 and Korea Composite Stock Price Index 200 (KOSPI200)). The results confirm the excellent forecasting accuracy of the proposed model. ModAugNet-c yields a lower test error than the comparative model (SingleNet) in which an overfitting prevention LSTM module is absent. The test mean squared error (MSE), mean absolute percentage error (MAPE), and mean absolute error (MAE) for S&P500 decreased to 54.1%, 35.5%, and 32.7%, respectively, of the corresponding S&P500 forecasting errors of SingleNet, while the same for KOSPI200 decreased to 48%, 23.9%, and 32.7%, respectively, of the corresponding KOSPI200 forecasting errors of SingleNet. Furthermore, through the analyses of the trained ModAugNet-c, we found that test performance is entirely dependent on the prediction LSTM module.
The contribution of this study is its applicability in various instances where it is challenging to artificially augment data, such as medical data analysis and financial time-series modeling.
•A financial trading system is proposed to improve traders’ profits.•The system uses the number of shares, action strategies, and transfer learning.•The number of shares is determined by using a DNN ...regressor.•When confusion exists, postponing a financial decision is the best policy.•Transfer learning can address problems of insufficient financial data.
We study trading systems using reinforcement learning with three newly proposed methods to maximize total profits and reflect real financial market situations while overcoming the limitations of financial data. First, we propose a trading system that can predict the number of shares to trade. Specifically, we design an automated system that predicts the number of shares by adding a deep neural network (DNN) regressor to a deep Q-network, thereby combining reinforcement learning and a DNN. Second, we study various action strategies that use Q-values to analyze which action strategies are beneficial for profits in a confused market. Finally, we propose transfer learning approaches to prevent overfitting from insufficient financial data. We use four different stock indices—the S&P500, KOSPI, HSI, and EuroStoxx50—to experimentally verify our proposed methods and then conduct extensive research. The proposed automated trading system, which enables us to predict the number of shares with the DNN regressor, increases total profits by four times in S&P500, five times in KOSPI, 12 times in HSI, and six times in EuroStoxx50 compared with the fixed-number trading system. When the market situation is confused, delaying the decision to buy or sell increases total profits by 18% in S&P500, 24% in KOSPI, and 49% in EuroStoxx50. Further, transfer learning increases total profits by twofold in S&P500, 3 times in KOSPI, twofold in HSI, and 2.5 times in EuroStoxx50. The trading system with all three proposed methods increases total profits by 13 times in S&P500, 24 times in KOSPI, 30 times in HSI, and 18 times in EuroStoxx50, outperforming the market and the reinforcement learning model.
•A new hybrid method to integrate deep neural networks with multiple financial time series models is proposed.•Combines the LSTM model with various generalized autoregressive conditional ...heteroskedasticity (GARCH)-type models.•Compared performance of the proposed hybrid LSTM models with that of existing methodologies.
Volatility plays crucial roles in financial markets, such as in derivative pricing, portfolio risk management, and hedging strategies. Therefore, accurate prediction of volatility is critical. We propose a new hybrid long short-term memory (LSTM) model to forecast stock price volatility that combines the LSTM model with various generalized autoregressive conditional heteroscedasticity (GARCH)-type models. We use KOSPI 200 index data to discover proposed hybrid models that combine an LSTM with one to three GARCH-type models. In addition, we compare their performance with existing methodologies by analyzing single models, such as the GARCH, exponential GARCH, exponentially weighted moving average, a deep feedforward neural network (DFN), and the LSTM, as well as the hybrid DFN models combining a DFN with one GARCH-type model. Their performance is compared with that of the proposed hybrid LSTM models. We discover that GEW-LSTM, a proposed hybrid model combining the LSTM model with three GARCH-type models, has the lowest prediction errors in terms of mean absolute error (MAE), mean squared error (MSE), heteroscedasticity adjusted MAE (HMAE), and heteroscedasticity adjusted MSE (HMSE). The MAE of GEW-LSTM is 0.0107, which is 37.2% less than that of the E-DFN (0.017), the model combining EGARCH and DFN and the best model among those existing. In addition, the GEW-LSTM has 57.3%, 24.7%, and 48% smaller MSE, HMAE, and HMSE, respectively. The first contribution of this study is its hybrid LSTM model that combines excellent sequential pattern learning with improved prediction performance in stock market volatility. Second, our proposed model markedly enhances prediction performance of the existing literature by combining a neural network model with multiple econometric models rather than only a single econometric model. Finally, the proposed methodology can be extended to various fields as an integrated model combining time-series and neural network models as well as forecasting stock market volatility.
A general circulation model is used to study the interaction between parameterized gravity waves (GWs) and large‐scale Kelvin waves in the tropical stratosphere. The simulation shows that Kelvin ...waves with substantial amplitudes (∼10 m s−1) can significantly affect the distribution of GW drag by modulating the local shear. Furthermore, this effect is localized to regions above strong convective organizations that generate large‐amplitude GWs, so that at a given altitude it occurs selectively in a certain phase of Kelvin waves. Accordingly, this effect also contributes to the zonal‐mean GW drag, which is large in the middle stratosphere during the phase transition of the quasi‐biennial oscillation (QBO). Furthermore, we detect an enhancement of Kelvin‐wave momentum flux due to GW drag modulated by Kelvin waves. The result implies an importance of GW dynamics coupled to Kelvin waves in the QBO progression.
Plain Language Summary
The variability of the tropical atmosphere at altitudes of about 18–40 km is dominated by a large‐amplitude long‐term oscillation of wind, the quasi‐biennial oscillation, which has a broad impact on the climate and seasonal forecasting. This oscillation is known to be driven by various types of atmospheric waves with multiple spatial scales. Using a numerical model, this study reports a process of interaction between those waves on different scales, which has not been illuminated before. The result implies a potential importance of this process in the progression of the quasi‐biennial oscillation. Proper model representations of these multiscale waves and tropical convection are required to simulate this process.
Key Points
Kelvin waves affect the longitudinal and vertical distribution of parameterized gravity‐wave drag in the stratosphere significantly
This effect can make contribution to the zonal mean of gravity‐wave drag, thereby affecting the quasi‐biennial oscillation progression
Gravity‐wave drag modulated by Kelvin waves also alters the Kelvin‐wave momentum flux in the middle stratosphere
This study elucidates the mechanism behind persulfate activation by exploring the role of various oxyanions (e.g., peroxymonosulfate, periodate, and peracetate) in two activation systems utilizing ...iron nanoparticle (nFe0) as the reducing agent and single-wall carbon nanotubes (CNTs) as electron transfer mediators. Since the tested oxyanions serve as both electron acceptors and radical precursors in most cases, oxidative degradation of organics was achievable through one-electron reduction of oxyanions on nFe0 (leading to radical-induced oxidation) and electron transfer mediation from organics to oxyanions on CNTs (leading to oxidative decomposition involving no radical formation). A distinction between degradative reaction mechanisms of the nFe0/oxyanion and CNT/oxyanion systems was made in terms of the oxyanion consumption efficacy, radical scavenging effect, and EPR spectral analysis. Statistical study of substrate-specificity and product distribution implied that the reaction route induced on nFe0 varies depending on the oxyanion (i.e., oxyanion-derived radical), whereas the similar reaction pathway initiates organic oxidation in the CNT/oxyanion system irrespective of the oxyanion type. Chronoamperometric measurements further confirmed electron transfer from organics to oxyanions in the presence of CNTs, which was not observed when applying nFe0 instead.
Abstract
Background
Cytoplasmic inclusions of transactive response DNA binding protein of 43 kDa (TDP-43) in neurons and astrocytes are a feature of some neurodegenerative diseases, such as ...frontotemporal lobar degeneration with TDP-43 (FTLD-TDP) and amyotrophic lateral sclerosis (ALS). However, the role of TDP-43 in astrocyte pathology remains largely unknown.
Methods
To investigate whether TDP-43 overexpression in primary astrocytes could induce inflammation, we transfected primary astrocytes with plasmids encoding
Gfp
or
TDP
-
43
-
Gfp
. The inflammatory response and upregulation of PTP1B in transfected cells were examined using quantitative RT-PCR and immunoblot analysis. Neurotoxicity was analysed in a transwell coculture system of primary cortical neurons with astrocytes and cultured neurons treated with astrocyte-conditioned medium (ACM). We also examined the lifespan, performed climbing assays and analysed immunohistochemical data in pan-glial TDP-43-expressing flies in the presence or absence of a
Ptp61f
RNAi transgene.
Results
PTP1B inhibition suppressed TDP-43-induced secretion of inflammatory cytokines (interleukin 1 beta (IL-1β), interleukin 6 (IL-6) and tumour necrosis factor alpha (TNF-α)) in primary astrocytes. Using a neuron-astrocyte coculture system and astrocyte-conditioned media treatment, we demonstrated that PTP1B inhibition attenuated neuronal death and mitochondrial dysfunction caused by overexpression of TDP-43 in astrocytes. In addition, neuromuscular junction (NMJ) defects, a shortened lifespan, inflammation and climbing defects caused by pan-glial overexpression of TDP-43 were significantly rescued by downregulation of
ptp61f
(the
Drosophila
homologue of PTP1B) in flies.
Conclusions
These results indicate that PTP1B inhibition mitigates the neuronal toxicity caused by TDP-43-induced inflammation in mammalian astrocytes and
Drosophila
glial cells.
Cyclin‐dependent kinase 12 (CDK12) has emerged as an effective therapeutic target due to its ability to regulate DNA damage repair in human cancers, but little is known about the role of CDK12 in ...driving tumorigenesis. Here, we demonstrate that CDK12 promotes tumor initiation as a novel regulator of cancer stem cells (CSCs) and induces anti‐HER2 therapy resistance in human breast cancer. High CDK12 expression caused by concurrent amplification of CDK12 and HER2 in breast cancer patients is associated with disease recurrence and poor survival. CDK12 induces self‐renewal of breast CSCs and in vivo tumor‐initiating ability, and also reduces susceptibility to trastuzumab. Furthermore, CDK12 kinase activity inhibition facilitates anticancer efficacy of trastuzumab in HER2+ tumors, and mice bearing trastuzumab‐resistant HER2+ tumor show sensitivity to an inhibitor of CDK12. Mechanistically, the catalytic activity of CDK12 is required for the expression of genes involved in the activation of ErbB‐PI3K‐AKT or WNT‐signaling cascades. These results suggest that CDK12 is a major oncogenic driver and an actionable target for HER2+ breast cancer to replace or augment current anti‐HER2 therapies.
Synopsis
CDK12, a RNA polymerase II kinase, promotes breast cancer stem cell‐like properties mediated by WNT and ErbB‐PI3K signaling. Targeting CDK12 improves trastuzumab therapy in breast cancers characterised by HER2/CDK12 co‐amplification.
CDK12 amplification induces tumor initiation and progression, and mediates trastuzumab resistance.
CDK12 is required for transcriptional upregulation of genes involved in ErbB‐PI3K and WNT pathway activation.
CDK12 inhibition alone or in combination with trastuzumab therapy has anti‐tumor activity against HER2+ breast cancers.
CDK12, a RNA polymerase II kinase, promotes breast cancer stem cell‐like properties mediated by WNT and ErbB‐PI3K signaling. Targeting CDK12 improves trastuzumab therapy in breast cancers characterised by HER2/CDK12 co‐amplification.
Chloroplast genomes are valuable for inferring evolutionary relationships. We report the complete chloroplast genomes of 36 Corydalis spp. and one Fumaria species. We compared these genomes with 22 ...other taxa and investigated the genome structure, gene content, and evolutionary dynamics of the chloroplast genomes of 58 species, explored the structure, size, repeat sequences, and divergent hotspots of these genomes, conducted phylogenetic analysis, and identified nine types of chloroplast genome structures among Corydalis spp. The ndh gene family suffered inversion and rearrangement or was lost or pseudogenized throughout the chloroplast genomes of various Corydalis species. Analysis of five protein-coding genes revealed simple sequence repeats and repetitive sequences that can be potential molecular markers for species identification. Phylogenetic analysis revealed three subgenera in Corydalis. Subgenera Cremnocapnos and Sophorocapnos represented the Type 2 and 3 genome structures, respectively. Subgenus Corydalis included all types except type 3, suggesting that chloroplast genome structural diversity increased during its differentiation. Despite the explosive diversification of this subgenus, most endemic species collected from the Korean Peninsula shared only one type of genome structure, suggesting recent divergence. These findings will greatly improve our understanding of the chloroplast genome of Corydalis and may help develop effective molecular markers.