Passively harmonic mode-locked erbium-doped fiber laser at a 580 MHz repetition rate based on carbon nanotubes film

We propose and demonstrate a passively harmonic mode-locked erbium-doped fiber laser（EDFL） using carbon nanotubes polyvinyl alcohol（CNTs-PVA） film. The laser allows generation of the pulses with a repetition rate of 580 MHz, which corresponds to the 22nd harmonics of a 26.3 MHz fundamental repetition rate under 323 mW pump power. A particularly noteworthy feature of the pulses is the super-mode suppression ratio（SMSR）, which is over 40 dB, indicating a stable operation.

Corporate Probability of Default(PD)Modelling for Banks in Emerging Economies:A Case Study of Zimbabwe Stock Exchange(ZSE)Listed Counters

The paper extends Merton’s Probability of Default(PD)model to the case for transaction costs or market friction for estimation of the PDs of listed banking corporations.A closed form formula for the PD model is obtained and validated using financial data drawn from banks listed on the Zimbabwe Stock Exchange(ZSE).It has been observed that most corporations in emerging economies have been finding it extremely difficult to list,continue listed or manage risk emanating from credit exposures undertaken.In the absence of risk the role of the financial sector of an economy to efficiently and effectively allocate resources between the public and private sectors would be simplified,economically and rationally determined.Reliable or precise computation of the Probability of Default(PD)of a borrower is one of the most critical tasks in credit risk management for commercial banks that were applying the Internal Rating Based Approach(IRBA)under the Basel Capital Accords Ⅱ and Ⅲ frameworks.The study sought to develop a Probability of Default(PD)model that banking corporations in emerging economies such as Zimbabwe could adopt and implement in the Multiple Currency System(MCS)in their desire to grow and develop through their lending businesses.The research study adopted a PD model similar to the Asset Valuation Model(AVM)by Merton(1974)and initially extended by Black-Scholes(1973)and Crouhy et al.(2000)and applied it on a basket of Zimbabwe Stock Exchange listed counters after having adjusted the model for the transaction cost variable.The study therefore succeeded in coming up with a PD model that was worth adopting and implementing by Zimbabwe Stock Exchange(ZSE)listed corporations in their desire to grow towards sustainable development.It was realised that a contemporary PD model adjusted for transaction cost is pertinent for reflection of practical conditions banks face in estimation of their risk metrics such as PD.Transaction costs faced by banks in emerging economies are very huge that they cannot be assumed to be insignificant when it comes to valuation of PDs of banking corporations.The inclusion of transaction costs in estimation of PDs of ZSE listed banks is likely to create a paradigm shift in financial theory on risk metrics in the modern world.The study ends by recommending the need for all Zimbabwean listed corporations to adopt and implement an AVM adjusted for transaction costs if they were to successfully measure and manage both their investment and credit exposure endeavours in the multiple currency system period.

Profile and Rating Similarity Analysis for Recommendation Systems Using Deep Learning

Recommendation systems are going to be an integral part of any E-Business in near future.As in any other E-business,recommendation systems also play a key role in the travel business where the user has to be recommended with a restaurant that best suits him.In general,the recommendations to a user are made based on similarity that exists between the intended user and the other users.This similarity can be calculated either based on the similarity between the user profiles or the similarity between the ratings made by the users.First phase of this work concentrates on experimentally analyzing both these models and get a deep insight of these models.With the lessons learned from the insights,second phase of the work concentrates on developing a deep learning model.The model does not depend on the other user's profile or rating made by them.The model is tested with a small restaurant dataset and the model can predict whether a user likes the restaurant or not.The model is trained with different users and their rating.The system learns from it and in order to predict whether a new user likes or not a restaurant that he/she has not visited earlier,all the data the trained model needed is the rating made by the same user for different restaurants.The model is deployed in a cloud environment in order to extend it to be more realistic product in future.Result evaluated with dataset,it achieves 74.6%is accurate prediction of results,where as existing techniques achieves only 64%.

Measuring the model risk-adjusted performance of machine learning algorithms in credit default prediction

Implementing new machine learning(ML)algorithms for credit default prediction is associated with better predictive performance;however,it also generates new model risks,particularly concerning the supervisory validation process.Recent industry surveys often mention that uncertainty about how supervisors might assess these risks could be a barrier to innovation.In this study,we propose a new framework to quantify model risk-adjustments to compare the performance of several ML methods.To address this challenge,we first harness the internal ratings-based approach to identify up to 13 risk components that we classify into 3 main categories—statistics,technology,and market conduct.Second,to evaluate the importance of each risk category,we collect a series of regulatory documents related to three potential use cases—regulatory capital,credit scoring,or provisioning—and we compute the weight of each category according to the intensity of their mentions,using natural language processing and a risk terminology based on expert knowledge.Finally,we test our framework using popular ML models in credit risk,and a publicly available database,to quantify some proxies of a subset of risk factors that we deem representative.We measure the statistical risk according to the number of hyperparameters and the stability of the predictions.The technological risk is assessed through the transparency of the algorithm and the latency of the ML training method,while the market conduct risk is quantified by the time it takes to run a post hoc technique(SHapley Additive exPlanations)to interpret the output.

Empirical Research on Brand Spillover Effect of Agricultural Product under the Background of Product Injury Crisis

By referring to attribution theory,influence factor and internal mechanism of crisis brand generating two kinds of spillover effects are explored. In this paper,the influence mechanism on spillover effect is explored from the angles of network and industry information. Results show that high online review dispersion is easy to cause that consumer attributes the responsibility to the outside of the enterprise,while low online review dispersion is easy to cause that consumer attributes the responsibility to the inside of the enterprise; high industry base rate information is easy to cause that consumer attributes the responsibility to the outside of the enterprise,while low industry base rate information is easy to cause that consumer attributes the responsibility to the inside of the enterprise. When consumer attributes the responsibility to the outside of the enterprise,infectious network spillover is easy to occur. When consumer attributes the responsibility to the inside of the enterprise,contrast network spillover is easy to occur. Consumer's existing brand knowledge regulates the influence of online review dispersion on responsibility attribution and the influence of industry base rate information on responsibility attribution. Research result could provide effective theoretic reference and management suggestion for crisis enterprise and even whole industry responding to product injury crisis.

Midinfrared optical frequency comb based on difference frequency generation using high repetition rate Er-doped fiber laser with single wall carbon nanotube film

We demonstrated stable midinfrared(MIR) optical frequency comb at the 3.0 μm region with difference frequency generation pumped by a high power, Er-doped, ultrashort pulse fiber laser system. A soliton mode-locked161 MHz high repetition rate fiber laser using a single wall carbon nanotube was fabricated. The output pulse was amplified in an Er-doped single mode fiber amplifier, and a 1.1–2.2 μm wideband supercontinuum(SC) with an average power of 205 m W was generated in highly nonlinear fiber. The spectrogram of the generated SC was examined both experimentally and numerically. The generated SC was focused into a nonlinear crystal, and stable generation of MIR comb around the 3 μm wavelength region was realized.

Application of new directional logic to improve DC side fault discrimination for high resistance faults in HVDC grids

This paper proposes a simple and fast way to determine the direction of a fault in a multi-terminal high voltage direct current(HVDC) grid by comparing the rate of change of voltage(ROCOV) values at either side of the di/dt limiting inductors at the line terminals. A local measurement based secure and fast protection method is implemented by supervising a basic ROCOV relay with a directional element. This directional information is also used to develop a slower communication based DC line protection scheme for detecting high resistance faults. The proposed protection scheme is applied to a multi-level modular converter based three-terminal HVDC grid and its security and sensitivity are evaluated through electromagnetic transient simulations. A methodology to set the protection thresholds considering the constraints imposed by the breaker technology and communication delays is also presented. With properly designed di/dt limiting inductors,the ability of clearing any DC transmission system fault before fault currents exceeds a given breaker capacity is demonstrated.