No abstract

No abstract

Pulsed near-infrared radiation has been proposed as an alternative stimulus for auditory nerve stimulation and could be potentially used in the design of cochlear implant. Although the infrared with high absorption coefficient of water (i.e., wavelength ranged from 1.8 to 2.2 μm) has been widely investigated, the lymph in the cochlea absorbs most of the infrared energies, and only a small part can arrive at the target auditory nerves. The present study is aimed to test whether the short-wavelength near-infrared irradiation with lower absorption coefficients can penetrate the lymph fluid to stimulate the auditory nerves. An 808-nm near-infrared laser was chosen to stimulate the auditory nerve in the guinea pig cochlea. The infrared pulse was delivered by an optical fiber that was surgically inserted near the round window membrane and oriented toward the spiral ganglion cells in the basal turn of the cochlea. The 2-Hz infrared pulses were used to stimulate the cochlea before and after the deafness with different pulse durations (100-1,000 μs). Optically evoked compound action potentials (oCAPs) were recorded during the infrared radiation. We successfully recorded oCAPs from both normal hearing animals and deafened animals. The oCAP amplitude increased with the infrared radiation energy. The preliminary experiment suggests that the near-infrared with lower absorption coefficients can effectively pass through the lymph filled in the cochlea and stimulate the auditory nerve. Further studies will optimize the deafness animal model and determine the optimal stimulation parameters.

Deep brain stimulation (DBS) is a neurosurgical technique that regulates neuron activity by using internal pulse generators to electrodes in specific target areas of the brain. As a blind treatment, DBS is widely used in the field of mental and neurological diseases, although its mechanism of action is still unclear. In the past 10 years, DBS has shown a certain positive effect in animal models and patients with Alzheimer's disease (AD), but there are also different results that may be related to the stimulation parameters of DBS. Based on this, determining the optimal stimulation parameters for DBS in AD and understanding its mechanism of action are essential to promote the clinical application of DBS in AD. This review aims to explore the therapeutic effect of DBS in AD, and to analyze its stimulation parameters and potential mechanism of action. The keywords “Deep brain stimulation” and “Alzheimer's Disease” were used for systematic searches in the literature databases of Web of Science and PubMed (from 1900 to September 29, 2020). All human clinical studies and animal studies were reported in English, including individual case studies and long-term follow-up studies, were included. These studies described the therapeutic effects of DBS in AD. The results included 16 human clinical studies and 14 animal studies, of which 28 studies clearly demonstrated the positive effect of DBS in AD. We analyzed the current stimulation parameters of DBS in AD from stimulation target, stimulation frequency, stimulation start time, stimulation duration, unilateral/bilateral treatment and current intensity, etc., and we also discussed its potential mechanism of action from multiple aspects, including regulating related neural networks, promoting nerve oscillation, reducing β-amyloid and tau levels, reducing neuroinflammation, regulating the cholinergic system, inducing the synthesis of nerve growth factor.

No abstract

No abstract

Multi-head attention is appealing for its ability to jointly extract different types of information from multiple representation subspaces. Concerning the information aggregation, a common practice is to use a concatenation followed by a linear transformation, which may not fully exploit the expressiveness of multi-head attention. In this work, we propose to improve the information aggregation for multi-head attention with a more powerful routing-by-agreement algorithm. Specifically, the routing algorithm iteratively updates the proportion of how much a part (i.e. the distinct information learned from a specific subspace) should be assigned to a whole (i.e. the final output representation), based on the agreement between parts and wholes. Experimental results on linguistic probing tasks and machine translation tasks prove the superiority of the advanced information aggregation over the standard linear transformation.

Managing uncertainty has been a challenging task for market operations. This paper first reviews the current practice of managing uncertainties at MISO. A framework of using robust optimization based approach on MISO Look-Ahead commitment (LAC) is then introduced. The numerical results show that this type of approaches are promising and yet with challenges to overcome in order to be practical for real world application.