HyperNOMAD

The performance of deep neural networks is highly sensitive to the choice of the hyperparameters that define the structure of the network and the learning process. When facing a new application, tuning a deep neural network is a tedious and time-consuming process that is often described as a “dark art.” This explains the necessity of automating the calibration of these hyperparameters. Derivative-free optimization is a field that develops methods designed to optimize time-consuming functions without relying on derivatives. This work introduces the HyperNOMAD package, an extension of the NOMAD software that applies the MADS algorithm [7] to simultaneously tune the hyperparameters responsible for both the architecture and the learning process of a deep neural network (DNN). This generic approach allows for an important flexibility in the exploration of the search space by taking advantage of categorical variables. HyperNOMAD is tested on the MNIST, Fashion-MNIST, and CIFAR-10 datasets and achieves results comparable to the current state of the art.

Cyanidin-3-glucoside Lipophilic Conjugates for Topical Application: Tuning the Antimicrobial Activities with Fatty Acid Chain Length

Background: Natural anthocyanins present a low solubility in lipophilic media, which compromises their effective application in lipophilic systems. In this work, cyanidin-3-O-glucoside (Cy3glc) was esterified by the addition of fatty acids with increasing chain-lengths and a structure-activity relationship was performed towards the description of the best analog for skin-care applications. Methods: By enzymatic hemi-synthesis, it was possible to obtain 5 structurally related derivatives of cyanidin-3-O-glucoside with successive C2 increments in the aliphatic chain. The stability in hanks buffer and DMEM with or without FBS was followed by HPLC. The cytotoxicity against keratinocytes was evaluated by MTT assay. The antioxidant capacity was determined by using the fluorescent probe DCF-DA. The effect on enzyme activity was evaluated towards tyrosinase, collagenase, and elastase enzymes by colorimetric assays. MIC and MBC values were obtained against reference strains and against multidrug-resistant isolates. Results: In physiological conditions, cy3glc−fatty acid derivatives are more stable and may be converted to the native anthocyanin. The 5 conjugates showed lower antioxidant capacity and enzymatic inhibitory activities in comparison to the anthocyanin precursor. However, concerning the antibacterial activity, the insertion of a fatty acid chain sprouted the antibacterial activity, showing a clear biphasic effect and a more effective effect on Gram-positive bacteria. Conclusions: Cy3glc-C10 was the most effective compound considering the antimicrobial activity, although a general reduction was observed among the other activities evaluated. This work prompt further assays with a different panoply of derivatives ranging other features including saturation vs. unsaturation, even vs. odd carbon content and linear vs. branched.

Nanostructuring few-layer graphene films with swift heavy ions for electronic application: tuning of electronic and transport properties

The morphology and electronic properties of single and few-layer graphene films nanostructured by the impact of heavy high-energy ions have been studied.

SEGMENTASI KOLEKSI DATA UNTUK MENDUKUNG PERFORMANSI SISTEM

[Id] Pengelolaan koleksi data merupakan hal yang sangat kritis, apalagi jika melibatkan volume data yang sangat besar, karena hal ini akan berpengaruh terhadap performansi sistem secara menyeluruh. Suatu rancangan strategi khusus dibutuhkan bagi penyediaan performansi sistem, sehingga sistem mampu mengerjakan tugas-tugasnya sesuai dengan indikator sistem. Salah satu inini. Makalah ini akan menguraikan salah satu persoalan yang berhubungan dengan application tuning, terutama beberapa hal yang berhubungan dengan kebutuhan penggunaan segmentasi atau partisi untuk mengelompokan koleksi data yang besar. Pembahasan diawali dengan latar belakang, dilanjutkan dengan teori-teori terkait, kemudian mendefinisikan ilustrasi kasus, analisis dan pembahasan kasus, proses implementasi dan pengujian, serta kesimpulan dan pekerjaan kedepan [EN] Management of data collection is one of critical matter, event if involving a large volume of data because its will affect to all system performance. a particular strategic planning is needed to provision system performance, then the system able to do its jobs suitable with system indicator. one of very general indicator that can be used to assess system performance is speed. Basically, speed connected with tuning technique, and at the current time, there are many types, both theory or methodology that can be used in tuning technique. This paper will depict one problem that related to tuning application, especially some item that related with the requirement of using segmentation or partition to grouping large data collection. Depiction started with a background, then continued with related theories, case illustration, analyze and case description, implementation process, testing, conclusion and future works.