Algorithms for parameter optimization problems of nonlinear discrete systems

Machine learning techniques lend themselves as promising decision-making and analytic tools in a wide range of applications. Different ML algorithms have various hyper-parameters. In order to tailor an ML model towards a specific application, a large number of hyper-parameters should be tuned. Tuning the hyper-parameters directly affects the performance (accuracy and run-time). However, for large-scale search spaces, efficiently exploring the ample number of combinations of hyper-parameters is computationally challenging. Existing automated hyper-parameter tuning techniques suffer from high time complexity. In this paper, we propose HyP-ABC, an automatic innovative hybrid hyper-parameter optimization algorithm using the modified artificial bee colony approach, to measure the classification accuracy of three ML algorithms, namely random forest, extreme gradient boosting, and support vector machine. Compared to the state-of-the-art techniques, HyP-ABC is more efficient and has a limited number of parameters to be tuned, making it worthwhile for real-world hyper-parameter optimization problems. We further compare our proposed HyP-ABC algorithm with state-of-the-art techniques. In order to ensure the robustness of the proposed method, the algorithm takes a wide range of feasible hyper-parameter values, and is tested using a real-world educational dataset.

Download Full-text

Necessary and sufficient conditions for oscillation of coupled nonlinear discrete systems

Difference Equations and Discrete Dynamical Systems ◽

10.1142/9789812701572_0012 ◽

2005 ◽

Author(s):

Serena Matucci ◽

Pavel Řehák

Keyword(s):

Sufficient Conditions ◽

Discrete Systems ◽

Necessary And Sufficient Conditions ◽

Necessary And Sufficient ◽

Nonlinear Discrete Systems

Download Full-text

Characterization of the Transient Response of Coupled Optimization in Multidisciplinary Design

Mathematical Problems in Engineering ◽

10.1155/2013/910209 ◽

2013 ◽

Vol 2013 ◽

pp. 1-15 ◽

Cited By ~ 1

Author(s):

Erich Devendorf ◽

Kemper Lewis

Keyword(s):

Optimization Problems ◽

Initial Conditions ◽

Systems Design ◽

Discrete Systems ◽

Product Development Process ◽

Multidisciplinary Design ◽

Distributed Design ◽

Equilibrium Solutions ◽

Acceptable Solution ◽

Number Of Iterations

Time is an asset of critical importance in a multidisciplinary design process and it is desirable to reduce the amount of time spent designing products and systems. Design is an iterative activity and designers consume a significant portion of the product development process negotiating a mutually acceptable solution. The amount of time necessary to complete a design depends on the number and duration of design iterations. This paper focuses on accurately characterizing the number of iterations required for designers to converge to an equilibrium solution in distributed design processes. In distributed design, systems are decomposed into smaller, coupled design problems where individual designers have control over local design decisions and seek to achieve their own individual objectives. These smaller coupled design optimization problems can be modeled using coupled games and the number of iterations required to reach equilibrium solutions varies based on initial conditions and process architecture. In this paper, we leverage concepts from game theory, classical controls, and discrete systems theory to evaluate and approximate process architectures without carrying out any solution iterations. As a result, we develop an analogy between discrete decisions and a continuous time representation that we analyze using control theoretic techniques.

Download Full-text