HYPERPARAMETER OPTIMIZATION BASED ON APRIORI AND A POSTERIORI KNOWLEDGE ABOUT CLASSIFICATION PROBLEM
Annotation
Subject of Research.The paper deals with Bayesian method for hyperparameter optimization of algorithms, used in machine learning for classification problems. A comprehensive survey is carried out about using a priori and a posteriori knowledge in classification task for hyperparameter optimization quality improvement. Method. The existing Bayesian optimization algorithm for hyperparameter setting in classification problems was expanded. We proposed a target function modification calculated on the basis of hyperparameters optimized for the similar problems and a metric for determination of similarity classification problems based on generated meta-features. Main Results. Experiments carried out on the real-world datasets from OpenML database have confirmed that the proposed algorithm achieves usually significantly better performance results than the existing Bayesian optimization algorithm within a fixed time limit.Practical Relevance. The proposed algorithm can be used for hyperparameter optimization in any classification problem, for example, in medicine, image processing or chemistry.
Keywords
Постоянный URL
Articles in current issue
- APPLICATION OF INDUCED MECHANICAL STRESSES IN FORMATION OF SPHERICAL SURFACES OF INTERFERENCE MIRROR SUBSTRATES
SENSITIVITY VARIATION RESEARCH OF TILTED FIBER BRAGG GRATING DURING CHEMICAL ETCHING
- APLICATION FEATURES OF OPTICAL POLYMERS IN OPTICAL SYSTEMS DESIGN
- EFFECT OF PLANT EXTRACTS ON ACTIVITY OF STAPHYLOCOCCUS AUREUS BY ELECTROCHEMICAL BIOTESTING
- FAST MOTION ESTIMATION ALGORITHM FOR HEVC VIDEO CODEC
- METHOD OF JOINT CLUSTERING IN NETWORK AND CORRELATION SPACES
- CONCEPT OF DIGITAL TWINS AT LIFE CYCLE STAGES OF PRODUCTION SYSTEMS
- INFERRING OF REGULATORY NETWORKS FROM EXPRESSION DATA USING BAYESIAN NETWORKS
- APPLICATION OF INCREMENTAL SATISFIABILITY PROBLEM SOLVERS FOR NON-DETERMINISTIC POLYNOMIAL-TIME HARD PROBLEMS AS ILLUSTRATED BY MINIMAL BOOLEAN FORMULA SYNTHESIS PROBLEM
- DISTRIBUTION EVALUATION OF REFLECTIVE CHARACTERISTICS WITH QUASI-CONTINUOUS ULTRA-WIDEBAND PROBING SIGNAL
- ADAPTIVE PROBLEM OF EXTENDED REPRODUCTION WITH MINIMIZATION OF GENERALIZED COSTS
- PREDICTION OF REACTION CONDITIONS BY DEEP LEARNING TECHNIQUES
- MODELING OF VERTICAL LIGHT PIPES FOR DAYLIGHT ILLUMINATION OF INDOOR INDUSTRIAL BUILDINGS(in English)
- CONDITION EQUATION OF POLYMER FILAMENTS
- PREDICTION OF MECHANICAL PROPERTIES FOR ONE-DIMENSIONAL POLYMER STRUCTURES
- INDEPENDENT COMPONENT ANALYSIS FOR INITIAL APPROXIMATION DETERMINATION IN IDENTIFICATION OF ACTIVE MODULES IN BIOLOGICAL GRAPHS
- CLINICAL DECISIONSUPPORT SYSTEM WITH PROCESSING OF MULTIMODAL MEDICAL DATA FOR RADIOLOGIST EFFICIENCY IMPROVEMENT PRACTICE