-
To tune or not to tune? An Approach for Recommending Important Hyperparameters
Authors:
Mohamadjavad Bahmani,
Radwa El Shawi,
Nshan Potikyan,
Sherif Sakr
Abstract:
Novel technologies in automated machine learning ease the complexity of algorithm selection and hyperparameter optimization. Hyperparameters are important for machine learning models as they significantly influence the performance of machine learning models. Many optimization techniques have achieved notable success in hyperparameter tuning and surpassed the performance of human experts. However,…
▽ More
Novel technologies in automated machine learning ease the complexity of algorithm selection and hyperparameter optimization. Hyperparameters are important for machine learning models as they significantly influence the performance of machine learning models. Many optimization techniques have achieved notable success in hyperparameter tuning and surpassed the performance of human experts. However, depending on such techniques as blackbox algorithms can leave machine learning practitioners without insight into the relative importance of different hyperparameters. In this paper, we consider building the relationship between the performance of the machine learning models and their hyperparameters to discover the trend and gain insights, with empirical results based on six classifiers and 200 datasets. Our results enable users to decide whether it is worth conducting a possibly time-consuming tuning strategy, to focus on the most important hyperparameters, and to choose adequate hyperparameter spaces for tuning. The results of our experiments show that gradient boosting and Adaboost outperform other classifiers across 200 problems. However, they need tuning to boost their performance. Overall, the results obtained from this study provide a quantitative basis to focus efforts toward guided automated hyperparameter optimization and contribute toward the development of better-automated machine learning frameworks.
△ Less
Submitted 30 August, 2021;
originally announced August 2021.
-
Effect of Interfacial Defects on the Electronic Properties of MoS$_2$ Based Lateral T-H Heterophase Junctions
Authors:
Mohammad Bahmani,
Mahdi Ghorbani-Asl,
Thomas Frauenheim
Abstract:
The coexistence of semiconducting (2H) and metallic (1T) phases of MoS$_{2}$ monolayers have further pushed their strong potential for applications in the next generation of electronic devices based on the two-dimensional lateral heterojunctions. Structural defects have considerable effects on the properties of these 2D devices. In particular, the interfaces of two phases are often imperfect and m…
▽ More
The coexistence of semiconducting (2H) and metallic (1T) phases of MoS$_{2}$ monolayers have further pushed their strong potential for applications in the next generation of electronic devices based on the two-dimensional lateral heterojunctions. Structural defects have considerable effects on the properties of these 2D devices. In particular, the interfaces of two phases are often imperfect and may contain numerous vacancies created by phase engineering techniques, e.g. under the electron beam. Here, the transport behaviors of the heterojunctions in the existence of point defects are explored by means of first-principles calculations and non-equilibrium Green's function approach. While vacancies in semiconducting MoS$_{2}$ act as scattering centers, their presence at the interface improves the flow of the charge carriers. In the case of $V_{Mo}$, the current has been increased by two orders of magnitude in comparison to the perfect device. The enhancement of transmission was explained by changes in the electronic densities at the T-H interface, which open new transport channels for electron conduction.
△ Less
Submitted 8 August, 2021;
originally announced August 2021.
-
A Novel Approximate Hamming Weight Computing for Spiking Neural Networks: an FPGA Friendly Architecture
Authors:
Kaveh Akbarzadeh-Sherbaf,
Mikaeel Bahmani,
Danial Ghiaseddin,
Saeed Safari,
Abdol-Hossein Vahabie
Abstract:
Hamming weights of sparse and long binary vectors are important modules in many scientific applications, particularly in spiking neural networks that are of our interest. To improve both area and latency of their FPGA implementations, we propose a method inspired from synaptic transmission failure for exploiting FPGA lookup tables to compress long input vectors. To evaluate the effectiveness of th…
▽ More
Hamming weights of sparse and long binary vectors are important modules in many scientific applications, particularly in spiking neural networks that are of our interest. To improve both area and latency of their FPGA implementations, we propose a method inspired from synaptic transmission failure for exploiting FPGA lookup tables to compress long input vectors. To evaluate the effectiveness of this approach, we count the number of `1's of the compressed vector using a simple linear adder. We classify the compressors into shallow ones with up to two levels of lookup tables and deep ones with more than two levels. The architecture generated by this approach shows up to 82% and 35% reductions for different configurations of shallow compressors in area and latency respectively. Moreover, our simulation results show that calculating the Hamming weight of a 1024-bit vector of a spiking neural network by the use of only deep compressors preserves the chaotic behavior of the network while slightly impacts on the learning performance.
△ Less
Submitted 29 April, 2021;
originally announced April 2021.
-
A technique to study the elastic and inelastic interaction of quarkonium with hadrons using femtoscopic correlations
Authors:
Marzieh Bahmani,
Daniel Kikoła,
Leszek Kosarzewski
Abstract:
We present a method for the measurement of parameters of elastic and inelastic interactions of charmonium with hadrons. In this technique, we use femtoscopic analysis of charmonium-hadron correlations at low relative momentum and the Lednicky-Lyuboshitz analytical model to extract the interaction parameters. We argue that such a study is already feasible in the LHCb experiment at the LHC, and we d…
▽ More
We present a method for the measurement of parameters of elastic and inelastic interactions of charmonium with hadrons. In this technique, we use femtoscopic analysis of charmonium-hadron correlations at low relative momentum and the Lednicky-Lyuboshitz analytical model to extract the interaction parameters. We argue that such a study is already feasible in the LHCb experiment at the LHC, and we discuss the prospects for studies in the STAR at RHIC and other experiments at the LHC.
△ Less
Submitted 21 December, 2020;
originally announced December 2020.
-
Reversible tuning the optical properties of defective TMDs monolayers
Authors:
Mohammad Bahmani,
Michael Lorke,
Mahdi Faghihnasiri,
Thomas Frauenheim
Abstract:
Potential applications of monolayer of transition metal dichalcogenides (TMDs) in optoelectronic and flexible devices are under heavy investigation. Although TMDs monolayers are highly robust to external mechanical fields, their electronic structure is sensitive to compressive and tensile strain. Besides, intrinsic point defects are present in synthesized samples of these two dimensional (2D) mate…
▽ More
Potential applications of monolayer of transition metal dichalcogenides (TMDs) in optoelectronic and flexible devices are under heavy investigation. Although TMDs monolayers are highly robust to external mechanical fields, their electronic structure is sensitive to compressive and tensile strain. Besides, intrinsic point defects are present in synthesized samples of these two dimensional (2D) materials which leads to the modification of their electronic and optical properties. Presence of vacancy complexes leads to absorption with larger dipole matrix elements in comparison to the case of simple transition metal vacancies. Using first principles calculations, we scrutinize the effect of various strain situations on the absorption spectra of such defective monolayers and show that strain engineering allows for reversible tuning of the optical properties.
△ Less
Submitted 11 September, 2021; v1 submitted 2 October, 2020;
originally announced October 2020.
-
SAF: Simulated Annealing Fair Scheduling for Hadoop Yarn Clusters
Authors:
Mahsa Ghanavatinasab,
Mastaneh Bahmani,
Reza Azmi
Abstract:
Apache introduced YARN as the next generation of the Hadoop framework, providing resource management and a central platform to deliver consistent data governance tools across Hadoop clusters. Hadoop YARN supports multiple frameworks like MapReduce to process different types of data and works with different scheduling policies such as FIFO, Capacity, and Fair schedulers. DRF is the best option that…
▽ More
Apache introduced YARN as the next generation of the Hadoop framework, providing resource management and a central platform to deliver consistent data governance tools across Hadoop clusters. Hadoop YARN supports multiple frameworks like MapReduce to process different types of data and works with different scheduling policies such as FIFO, Capacity, and Fair schedulers. DRF is the best option that uses short-term, without considering history information, convergence to fairness for multi-type resource allocation. However, DRF performance is still not satisfying due to trade-offs between fairness and performance regarding resource utilization. To address this problem, we propose Simulated Annealing Fair scheduling, SAF, a long-term fair scheme in resource allocation to have fairness and excellent performance in terms of resource utilization and MakeSpan. We introduce a new parameter as entropy, which is an approach to indicates the disorder in the fairness of allocated resources of the whole cluster. We implemented SAF as a pluggable scheduler in Hadoop Yarn Cluster and evaluated it with standard MapReduce benchmarks in Yarn Scheduler Load Simulator (SLS) and CloudSim Plus simulation framework. Finally, the results of both simulation tools are evidence to prove our claim. Compared to DRF, SAF increases resource utilization of YARN clusters significantly and decreases MakeSpan to an appropriate level.
△ Less
Submitted 28 August, 2020;
originally announced August 2020.
-
f-Biderivations and Jordan biderivations of unital algebras with idempotents
Authors:
Mohammad Ali Bahmani,
Driss Bennis,
Hamid Reza Ebrahimi Vishki,
Brahim Fahid
Abstract:
The notion of f-derivations was introduced by Beidar and Fong to unify several kinds of linear maps including derivations, Lie derivations and Jordan derivations. In this paper we introduce the notion of f-biderivations as a natural "biderivation" counterpart of the notion of "f-derivations". We first show, under some conditions, that any f-derivation is a Jordan biderivation. Then, we turn to stu…
▽ More
The notion of f-derivations was introduced by Beidar and Fong to unify several kinds of linear maps including derivations, Lie derivations and Jordan derivations. In this paper we introduce the notion of f-biderivations as a natural "biderivation" counterpart of the notion of "f-derivations". We first show, under some conditions, that any f-derivation is a Jordan biderivation. Then, we turn to study f-biderivations of a unital algebra with an idempotent. Our second main result shows, under some conditions, that every Jordan biderivation can be written as a sum of a biderivation, an antibiderivation and an extremal biderivation. As a consequence we show that every Jordan biderivation on a triangular algebra is a biderivation.
△ Less
Submitted 25 January, 2020;
originally announced January 2020.
-
Electronic Properties of Defective MoS$_{2}$ Monolayers Subject to Mechanical Deformations: A First-Principles Approach
Authors:
Mohammad Bahmani,
Mahdi Faghihnasiri,
Michael Lorke,
Agnieszka-Beata Kuc,
Thomas Frauenheim
Abstract:
Monolayers (ML) of Group-6 transition-metal dichalcogenides (TMDs) are semiconducting two-dimensional materials with direct bandgap, showing promising applications in various fields of science and technology, such as nanoelectronics and optoelectronics. These monolayers can undergo strong elastic deformations, up to about 10\%, without any bond breaking. Moreover, the electronic structure and tran…
▽ More
Monolayers (ML) of Group-6 transition-metal dichalcogenides (TMDs) are semiconducting two-dimensional materials with direct bandgap, showing promising applications in various fields of science and technology, such as nanoelectronics and optoelectronics. These monolayers can undergo strong elastic deformations, up to about 10\%, without any bond breaking. Moreover, the electronic structure and transport properties, which define the performance of these TMDs monolayers in nanoelectronic devices, can be strongly affected by the presence of point defects, which are often present in the synthetic samples. Thus, it is important to understand both effects on the electronic properties of such monolayers. In this work, we have investigated the electronic structure and energetic properties of defective MoS$_{2}$ monolayers, as subject to various strains, using density functional theory simulations. Our results indicated that strain leads to strong modifications of the defect levels inside the bandgap and their orbital characteristics. Strain also splits the degenerate defect levels up to an amount of 450~meV, proposing novel applications.
△ Less
Submitted 12 December, 2019; v1 submitted 29 April, 2019;
originally announced April 2019.
-
Production of tau tau jj final states at the LHC and the TauSpinner algorithm: the spin-2 case
Authors:
M. Bahmani,
J. Kalinowski,
W. Kotlarski,
E. Richter-Was,
Z. Was
Abstract:
The TauSpinner algorithm is a tool that allows to modify the physics model of the Monte Carlo generated samples due to the changed assumptions of event production dynamics, but without the need of re-generating events. With the help of weights $τ$-lepton production or decay processes can be modified accordingly to a new physics model. In a recent paper a new version TauSpinner ver.2.0.0 has been p…
▽ More
The TauSpinner algorithm is a tool that allows to modify the physics model of the Monte Carlo generated samples due to the changed assumptions of event production dynamics, but without the need of re-generating events. With the help of weights $τ$-lepton production or decay processes can be modified accordingly to a new physics model. In a recent paper a new version TauSpinner ver.2.0.0 has been presented which includes a provision for introducing non-standard states and couplings and study their effects in the vector-boson-fusion processes by exploiting the spin correlations of $τ$-lepton pair decay products in processes where final states include also two hard jets. In the present paper we document how this can be achieved taking as an example the non-standard spin-2 state that couples to Standard Model particles and tree-level matrix elements with complete helicity information included for the parton-parton scattering amplitudes into a $τ$-lepton pair and two outgoing partons. This implementation is prepared as the external (user provided) routine for the TauSpinner algorithm. It exploits amplitudes generated by MadGraph5 and adopted to the TauSpinner algorithm format. Consistency tests of the implemented matrix elements, reweighting algorithm and numerical results for observables sensitive to $τ$ polarization are presented.
△ Less
Submitted 15 January, 2018; v1 submitted 11 August, 2017;
originally announced August 2017.
