Improving Low-Latency Predictions in Multi-Exit Neural Networks via Block-Dependent Losses.

As the size of a model increases, making predictions using deep neural networks (DNNs) is becoming more computationally expensive. Multi-exit neural network is one promising solution that can flexibly make anytime predictions via early exits, depending on the current test-time budget which may vary over time in practice (e.g., self-driving cars with dynamically changing speeds). However, the prediction performance at the earlier exits is generally much lower than the final exit, which becomes a critical issue in low-latency applications having a tight test-time budget. Compared to the previous works where each block is optimized to minimize the losses of all exits simultaneously, in this work, we propose a new method for training multi-exit neural networks by strategically imposing different objectives on individual blocks. The proposed idea based on grouping and overlapping strategies improves the prediction performance at the earlier exits while not degrading the performance of later ones, making our scheme to be more suitable for low-latency applications. Extensive experimental results on both image classification and semantic segmentation confirm the advantage of our approach. The proposed idea does not require any modifications in the model architecture and can be easily combined with existing strategies aiming to improve the performance of multi-exit neural networks.

Bronchitis, COPD, and pneumonia after viral endemic of patients with leprosy on Sorok Island in South Korea.

Viral respiratory diseases (VRDs) cause lung inflammation and inflammatory cytokine production. We study whether dapsone is responsible for its observed preventive treatment effects of the sustained viral RNA interferon response. Around 2008 and 2012, Korea's Dementia Management Act stipulated drastic changes in the administration of dementia medication by medical staff. Participants were randomized and we compared leprosy patients with VRDs after prescribing dapsone as a standard treatment from 2005 to 2019. Significance was evaluated based on the dapsone-prescribed (+) subgroup and the dapsone-unprescribed (-) subgroup of the VRD diagnosed (+) and VRD undiagnosed (-) subgroup. We analyzed VRD ( +)/(- with dapsone (+)/(-) group and used a T-test, and designed the equation of acetylation with dapsone and acetylcholine (AA) equation. The 6394 VRD participants who received the dapsone intervention compared to the 3255 VRD participants in the control group demonstrated at T2 VRD (+) dapsone (-) (mean (M) = 224.80, SD = 97.50): T3 VRD (-) dapsone (+) (M = 110.87, SD = 103.80), proving that VRD is low when dapsone is taken and high when it is not taken. The t value is 3.10, and the p value is 0.004395 (significant at p < 0.05). After an increase in VRDs peaked in 2009, bronchitis, COPD, and pneumonia surged in 2013. The AA equation was strongly negatively correlated with the prevalence of bronchitis and chronic obstructive pulmonary disease (COPD): with bronchitis, r(15) = -0.823189, p = 0.005519, and with COPD, r(15) = -0.8161, p = 0.000207 (significant at p < 0.05). Dapsone treated both bronchitis and COPD. This study provides theoretical clinical data to limit acetylcholine excess during the VRD pandemic for bronchitis, COPD, and pneumonia.

The Neuroinflammasome in Alzheimer's Disease and Cerebral Stroke.

AIM/BACKGROUND: This review investigated a patient with Alzheimer's disease (AD) treated with 4,4'-diaminodiphenyl sulfone (DDS) as a neuroinflammasome competitor. METHODS: We monitored AD's progression through numeric clinical staging (NCS) with a new biomarker. NCS was determined by the presence of AD symptoms and neuropsychiatric (NP) symptoms caused by anti-AD (AAD) drugs (D) as a biomarker. We also monitored the function of DDS for stroke in a no-intake emergency state. RESULTS: By introducing (D), AD's progression was monitored through NCS staging. AAD side effects and neuropsychiatric symptoms were identified. DDS was stopped in patients with stroke with NCS 6 caused by AAD, and it rapidly proceeded to cerebral infarct. CONCLUSIONS: AAD can occasionally exacerbate AD and stroke. DDS can alleviate mild cognitive impairment (MCI), early AD and stroke. We clinically confirmed the role of DDS as a neuroinflammasome competitor after stroke. DDS preserved neuronal survival within 24-55 h in the Seoul Study cohort.