Mohammad Amin

Mohammad Amin, Marta Molinas

Impedance measurements have been widely used with the Nyquist plot to estimate the stability of interconnected power systems. Being a black-box method for equivalent and aggregated impedance estimation, its use for the identification of sub-components bandwidth is not a straightforward task. This paper proposes a simple method that will enable to identify the specific part of the equivalent impedance (e.g. controller's bandwidth) that has major impact on the stability of the system. For doing that, the paper analyses the stability of an interconnected system of wind farms and high voltage dc (HVDC) transmission system. The impedance frequency responses of the wind farms and HVDC system from the ac collection point are measured and it is shown by the method proposed in this paper, which controller has major impact in the observed oscillation. A mitigation technique is proposed based on re-tuning of the critical controller bandwidth of the interconnected converters. The method suggested can reveal the internal controllers' dynamics of the wind farm from the measured impedance combined with an analytical expression of the impedance and a transfer function identity when no information about the controllers is provided by the vendors due to confidentiality and industry secrecy.

Mohammad Amin, Marta Molinas

Model Predictive Control (MPC) method is a class of advanced control techniques most widely applied in industry. The major advantages of the MPC are its straightforward procedure which can be applied for both linear and nonlinear system. This paper proposes the use of MPC for voltage source converter (VSC) in a high voltage direct current (HVDC) system. A MPC controller is modeled based on the state-space model of a single VSC-HVDC station including the dynamics of the main ac grid. A full scale nonlinear switching model of point-to-point connected VSC-based HVDC system is developed in Matlab/Simulink association with SimPower system to demonstrate the application of the proposed controller.

Hayato Kunugi, Mohsen Rahmani, Yosuke Iyama et al.

Deep generative modeling to stochastically design small molecules is an emerging technology for accelerating drug discovery and development. However, one major issue in molecular generative models is their lower frequency of drug-like compounds. To resolve this problem, we developed a novel framework for optimization of deep generative models integrated with a D-Wave quantum annealing computer, where our Neural Hash Function (NHF) presented herein is used both as the regularization and binarization schemes simultaneously, of which the latter is for transformation between continuous and discrete signals of the classical and quantum neural networks, respectively, in the error evaluation (i.e., objective) function. The compounds generated via the quantum-annealing generative models exhibited higher quality in both validity and drug-likeness than those generated via the fully-classical models, and was further indicated to exceed even the training data in terms of drug-likeness features, without any restraints and conditions to deliberately induce such an optimization. These results indicated an advantage of quantum annealing to aim at a stochastic generator integrated with our novel neural network architectures, for the extended performance of feature space sampling and extraction of characteristic features in drug design.

Sohini Roychowdhury, Adam Holeman, Mohammad Amin et al.

For online ad-recommendation systems, processing complete user-ad-engagement histories is both computationally intensive and noise-prone. We introduce Dynamix, a scalable, personalized sequence exploration framework that optimizes event history processing using maximum relevance principles and self-supervised learning through Event Based Features (EBFs). Dynamix categorizes users-engagements at session and surface-levels by leveraging correlations between dwell-times and ad-conversion events. This enables targeted, event-level feature removal and selective feature boosting for certain user-segments, thereby yielding training and inference efficiency wins without sacrificing engaging ad-prediction accuracy. While, dynamic resource removal increases training and inference throughput by 1.15% and 1.8%, respectively, dynamic feature boosting provides 0.033 NE gains while boosting inference QPS by 4.2% over baseline models. These results demonstrate that Dynamix achieves significant cost efficiency and performance improvements in online user-sequence based recommendation models. Self-supervised user-segmentation and resource exploration can further boost complex feature selection strategies while optimizing for workflow and compute resources.

Mohammad Amin, Farshad Safaei

Autism is one of the most important neurological disorders which leads to problems in a person's social interactions. Improvement of brain imaging technologies and techniques help us to build brain structural and functional networks. Finding networks topology pattern in each of the groups (autism and healthy control) can aid us to achieve an autism disorder screening model. In the present study, we have utilized the genetic algorithm to extract a discriminative sub-network that represents differences between two groups better. In the fitness evaluation phase, for each sub-network, a machine learning model was trained using various entropy features of the sub-network and its performance was measured. Proper model performance implies extracting a good discriminative sub-network. Network entropies can be used as network topological descriptors. The evaluation results indicate the acceptable performance of the proposed screening method based on extracted discriminative sub-networks and the machine learning models succeeded in obtaining a maximum accuracy of 73.1% in structural networks of the UCLA dataset, 82.2% in functional networks of the UCLA dataset, and 66.1% in functional networks of ABIDE datasets.