No abstract

Hyperparameter tuning is a critical function necessary for the effective deployment of most machine learning (ML) algorithms. It is used to find the optimal hyperparameter settings of an ML algorithm in order to improve its overall output performance. To this effect, several optimization strategies have been studied for fine-tuning the hyperparameters of many ML algorithms, especially in the absence of model-specific information. However, because most ML training procedures need a significant amount of computational time and memory, it is frequently necessary to build an optimization technique that converges within a small number of fitness evaluations. As a result, a simple deterministic selection genetic algorithm (SDSGA) is proposed in this article. The SDSGA was realized by ensuring that both chromosomes and their accompanying fitness values in the original genetic algorithm are selected in an elitist-like way. We assessed the SDSGA over a variety of mathematical test functions. It was then used to optimize the hyperparameters of two well-known machine learning models, namely, the convolutional neural network (CNN) and the random forest (RF) algorithm, with application on the MNIST and UCI classification datasets. The SDSGA’s efficiency was compared to that of the Bayesian Optimization (BO) and three other popular metaheuristic optimization algorithms (MOAs), namely, the genetic algorithm (GA), particle swarm optimization (PSO) and biogeography-based optimization (BBO) algorithms. The results obtained reveal that the SDSGA performed better than the other MOAs in solving 11 of the 17 known benchmark functions considered in our study. While optimizing the hyperparameters of the two ML models, it performed marginally better in terms of accuracy than the other methods while taking less time to compute.

Abstract-This paper presents a circuit-level model of a dualgate bilayer and four layer graphene field effect transistor (GFET). The model provides an accurate estimation of the conductance at the charge neutrality point (CNP). At the CNP the device has its maximum resistance, at which the model is validated against experimental data of the device off-current for a range of electric fields perpendicular to the channel. The model shows a good agreement for validations carried out at constant and varying temperatures. Using the general Schottky equation, the model estimates the amount of bandgap opening created by the application of an electric field. Also the model shows good agreement when validated against experiment for the channel output conductance against varying gate voltage for both a bilayer and four layer graphene channel.

This paper presents the concept of a bilayer graphene transistor using a floating gate to achieve the necessary threshold potential required for symmetrical transfer characteristics in complementary inverters. Using the charge injected into the floating-gate, the threshold voltage of the channel can be controlled. The control of the channel's electrostatic doping using a floating-gate is exploited to simulate an inverter which shows a symmetrical transfer characteristic centred at an input voltage of V dd /2.

This paper presents a hybrid Smell Agent Symbiosis Organism Search Algorithm (SASOS) for optimal control of autonomous microgrids. In microgrid operation, a single optimization algorithm often lacks the required balance between accuracy and speed to control power system parameters such as frequency and voltage effectively. The hybrid algorithm reduces the imbalance between exploitation and exploration and increases the effectiveness of control optimization in microgrids. To achieve this, various energy resource models were coordinated into a single model for optimal energy generation and distribution to loads. The optimization problem was formulated based on the network power flow and the discrete-time sampling of the constrained control parameters. The development of SASOS comprises components of Symbiotic Organism Search (SOS) and Smell Agent Optimization (SAO) codified in an optimization loop. Twenty-four standard test function benchmarks were used to evaluate the performance of the algorithm developed. The experimental analysis revealed that SASOS obtained 58.82% of the Desired Convergence Goal (DCG) in 17 of the benchmark functions. SASOS was implemented in the Microgrid Central Controller (MCC) and benchmarked alongside standard SOS and SAO optimization control strategies. The MATLAB/Simulink simulation results of the microgrid load disturbance rejection showed the viability of SASOS with an improved reduction in Total Harmonic Distortion (THD) of 19.76%, compared to the SOS, SAO, and MCC methods that have a THD reduction of 15.60%, 12.74%, and 6.04%, respectively, over the THD benchmark. Based on the results obtained, it can be concluded that SASOS demonstrates superior performance compared to other methods. This finding suggests that SASOS is a promising solution for enhancing the control system of autonomous microgrids. It was also shown to apply to other sectors of engineering optimization.

No abstract

Denial of service attack and its variants are the largest ravaging network problems. They are used to cause damage to network by disrupting its services in order to harm a business or organization. Flash event is a network phenomenon that causes surge in normal network flow due to sudden increase in number of network users, to curtail the menace of the Denial of service attack it is pertinent to expose the perpetrator and take appropriate action against it. Internet protocol traceback is a network forensic tool that is used to identify source of an Internet protocol packet. Most of presently available Internet protocol traceback tools that are based on bio-inspired algorithm employ flowbased search method for tracing source of a Denial of service attack without facility to differentiate flash event from the attack. Surge in network due to flash event can mislead such a traceback tool that uses flow-based search. This work presents a solution that uses hop-by-hop search with an incorporated discrimination policy implemented by shark smell optimization algorithm to differentiate the attack traffic from other traffics. It was tested on performance and convergence against an existing bio-inspired traceback tool that uses flowbase method and yielded outstanding results in all the tests.

No abstract