RESUMO
Volatility forecasting is a problem in finance that attracts the attention of both academia and industry. While existing approaches typically utilize a discrete-time latent process that governs the volatility to forecast its future level, volatility is considered to evolve continuously, which makes discrete-time modeling inevitably lose some critical information about the evolution of volatility. In this article, a novel neural-network-based model, Continuous Volatility Forecasting Model, CVFM is proposed to tackle this problem. First, CVFM introduces a continuous-time latent process, whose evolution is modeled with neural differential equations (NDEs), to govern volatility, which effectively captures the continuous evolutionary behavior of volatility in a data-driven way. Second, a scale-similarity-based mechanism is designed to calibrate the evolution equation of the latent process with real-world observations in the absence of high-frequency data. CVFM is tested on six real-world stock index datasets. The main experimental results show that CVFM can significantly outperform existing models in terms of both forecasting accuracy and high-volatility recognition.
RESUMO
Background: Previous studies have shown that the traditional Chinese medicine (TCM) called "compound healthy ear agent" (CHEA) had anti-apoptosis effects in cochlear hair cells and spiral ganglion neurons, and could protect mice hearing against presbycusis or age-related hearing loss (AHL), as well as aminoglycoside antibiotic-induced ototoxicity. Because its mechanisms of action are still unclear, we investigated the mechanism of action of CHEA against AHL in mice using proteomics techniques. Methods: Eighteen C57BL/6J mice at 1 month of age were randomly divided into three groups: (A) drinking water until 2 months of age, K2M); (B) drinking water until 7 months of age to induce AHL, K7M; (C) drinking water containing CHEA daily until 7 months of age as treatment group, Z7M. At 2 or 7 months mice were sacrificed and their cochleae were removed for proteomics analysis. Results: The numbers of proteins with a false discovery rate (FDR) < 1% were respectively 5873 for qualitative and 5492 for quantitative statistics. The numbers of proteins with differential enrichment at least 1.5-fold (p < 0.05) were respectively 351 for K7M vs K2M groups, 52 for Z7M vs K7M groups, 264 for Z7M vs K2M groups. The differentially expressed proteins in the Z7M group were involved in synaptic molecular transmission, energy metabolism, immune response, antioxidant defenses, and anti-apoptosis. Conclusion: The TCM CHEA played a protective role against AHL in mice by regulating the expression of specific proteins and genes in cochlear hair cells and spiral ganglion neurons. Besides the pathways expected to be involved (antioxidant and anti-apoptosis), proteins related to immune response is a new finding of the present study.