Search

Large Language Models (LLMs) such as ChatGPT possess advanced capabilities in understanding and generating text. These capabilities enable ChatGPT to create text based on specific instructions, which can serve as augmented data for text classification tasks. Previous studies have approached data augmentation (DA) by either rewriting the existing dataset with ChatGPT or generating entirely new data from scratch. However, it is unclear which method is better without comparing their effectiveness. This study investigates the application of both methods to two datasets: a general-topic dataset (Reuters news data) and a domain-specific dataset (Mitigation dataset). Our findings indicate that: 1. ChatGPT generated new data consistently enhanced model's classification results for both datasets. 2. Generating new data generally outperforms rewriting existing data, though crafting the prompts carefully is crucial to extract the most valuable information from ChatGPT, particularly for domain-specific data. 3. The augmentation data size affects the effectiveness of DA; however, we observed a plateau after incorporating 10 samples. 4. Combining the rewritten sample with new generated sample can potentially further improve the model's performance. [ABSTRACT FROM AUTHOR]

Endogenous formaldehyde (FA) is produced in the human body via various mechanisms to preserve healthy energy metabolism and safeguard the organism. However, endogenous FA can have several negative effects on the body through epigenetic alterations, including cancer growth promotion; neuronal, hippocampal and endothelial damages; atherosclerosis acceleration; haemopoietic stem cell destruction and haemopoietic cell production reduction. Certain medications with antioxidant effects, such as glutathione, vitamin E, resveratrol, alpha lipoic acid and polyphenols, lessen the detrimental effects of endogenous FA by reducing oxidative stress, directly scavenging endogenous FA or promoting its degradation. This study offers fresh perspectives for managing illnesses associated with endogenous FA exposure. This article reviews the physiological and pathological mechanisms underlying the production and metabolism of endogenous formaldehyde, as well as its bidirectional regulatory effects on the human body. Finally, drugs that eliminate endogenous formaldehyde are summarized. This review provides a new perspective on how to treat diseases caused by endogenous formaldehyde in the future. [ABSTRACT FROM AUTHOR]

Polar 4H-SiC nano-resonators can host localized surface phonon polaritons (SPhPs) with low optical loss and fully exploit light–matter interactions for promising nanophotonic applications such as surface-enhanced spectroscopies and thermal imaging. To expand the mid-infrared and infrared application spaces, a sound knowledge of temperature effect on infrared dielectric functions and SPhPs of 4H-SiC is required, yet it remains largely unexplored. Herein, we focus on exploiting the temperature influence on dielectric functions, hybridization of propagating and localized SPhPs, and tailed spectral radiation properties of 4H-SiC nanopillar arrays through spectroscopic ellipsometry (SE) measurements as well as multiscale simulations. The 4H-SiC crystal is grown using the physical vapor transport method, and SE experiments measure infrared dielectric functions at temperatures between 300 and 800 K. Finite-element electromagnetic simulations confirm the emerged Monopole and transverse dipoles (TD1 and TD2) resonance modes in 4H-SiC nano-resonators, which agrees with the literature experiment. At high temperatures with strong lattice vibration, the amplitudes of resonant optical absorption peaks gradually decrease and the linewidths broaden, accompanied by the weakened electric resonances. First-principles calculations show that the anharmonic phonon scattering strengthens and less optical phonons are coupled to incident photons as temperature increases. Moreover, the propagating and localized SPhPs' hybridization and spectral radiation properties of 4H-SiC nanopillar arrays can be largely tailed by modifying the morphology and incident angle of light. This work provides physical insights into the temperature-induced spectral tuning of 4H-SiC nano-resonators and helps exploit their applications in the high-temperature working conditions. [ABSTRACT FROM AUTHOR]