scholarly journals Phase Transitions in Transfer Learning for High-Dimensional Perceptrons

Entropy ◽  
2021 ◽  
Vol 23 (4) ◽  
pp. 400
Author(s):  
Oussama Dhifallah ◽  
Yue M. Lu
Keyword(s):  
Phase Transition ◽  
Phase Transitions ◽  
Transfer Learning ◽  
Negative Transfer ◽  
High Dimensional ◽  
Source Information ◽  
Generalization Performance ◽  
Target Task ◽  
Learning Tasks ◽  
Perceptron Learning

Transfer learning seeks to improve the generalization performance of a target task by exploiting the knowledge learned from a related source task. Central questions include deciding what information one should transfer and when transfer can be beneficial. The latter question is related to the so-called negative transfer phenomenon, where the transferred source information actually reduces the generalization performance of the target task. This happens when the two tasks are sufficiently dissimilar. In this paper, we present a theoretical analysis of transfer learning by studying a pair of related perceptron learning tasks. Despite the simplicity of our model, it reproduces several key phenomena observed in practice. Specifically, our asymptotic analysis reveals a phase transition from negative transfer to positive transfer as the similarity of the two tasks moves past a well-defined threshold.

scholarly journals Instance Transfer Learning with Multisource Dynamic TrAdaBoost

2014 ◽  
Vol 2014 ◽  
pp. 1-8 ◽  
Author(s):  
Qian Zhang ◽  
Haigang Li ◽  
Yong Zhang ◽  
Ming Li
Keyword(s):  
Transfer Learning ◽  
Classification Accuracy ◽  
Negative Transfer ◽  
Traditional Learning ◽  
Dynamic Factor ◽  
Similar Distribution ◽  
Target Domain ◽  
Learning Tasks ◽  
Sample Data ◽  
Better Than

Since the transfer learning can employ knowledge in relative domains to help the learning tasks in current target domain, compared with the traditional learning it shows the advantages of reducing the learning cost and improving the learning efficiency. Focused on the situation that sample data from the transfer source domain and the target domain have similar distribution, an instance transfer learning method based on multisource dynamic TrAdaBoost is proposed in this paper. In this method, knowledge from multiple source domains is used well to avoid negative transfer; furthermore, the information that is conducive to target task learning is obtained to train candidate classifiers. The theoretical analysis suggests that the proposed algorithm improves the capability that weight entropy drifts from source to target instances by means of adding the dynamic factor, and the classification effectiveness is better than single source transfer. Finally, experimental results show that the proposed algorithm has higher classification accuracy.

scholarly journals Electric properties of olive oil under pressure

2021 ◽  
Author(s):  
L. T. Pawlicki ◽  
R. M. Siegoczyński ◽  
S. Ptasznik ◽  
K. Marszałek
Keyword(s):  
Molecular Structure ◽  
Phase Transition ◽  
Phase Diagram ◽  
Phase Transitions ◽  
Olive Oil ◽  
Material Parameters ◽  
First Order ◽  
Long Time ◽  
Capacitive Reactance ◽  
First Order Phase

AbstractThe main purpose of the experiment was a thermodynamic research with use of the electric methods chosen. The substance examined was olive oil. The paper presents the resistance, capacitive reactance, relative permittivity and resistivity of olive. Compression was applied with two mean velocities up to 450 MPa. The results were shown as functions of pressure and time and depicted on the impedance phase diagram. The three first order phase transitions have been detected. All the changes in material parameters were observed during phase transitions. The material parameters measured turned out to be the much more sensitive long-time phase transition factors than temperature. The values of material parameters and their dependence on pressure and time were compared with the molecular structure, arrangement of molecules and interactions between them. Knowledge about olive oil parameters change with pressure and its phase transitions is very important for olive oil production and conservation.

Room-temperature NaI/H2O compression icing: solute–solute interactions

2017 ◽  
Vol 19 (39) ◽  
pp. 26645-26650 ◽  
Author(s):  
Qingxin Zeng ◽  
Chuang Yao ◽  
Kai Wang ◽  
Chang Q. Sun ◽  
Bo Zou
Keyword(s):  
Phase Transition ◽  
Phase Transitions ◽  
Bond Energy ◽  
Room Temperature ◽  
Solute Concentration ◽  
Solute Interaction ◽  
Solute Interactions

H–O bond energy governs the PCx for Na/H2O liquid–VI–VII phase transition. Solute concentration affects the path of phase transitions differently with the solute type. Solute–solute interaction lessens the PC2 sensitivity to compression. The PC1 goes along the liquid–VI boundary till the triple phase joint.

scholarly journals The origin of the lattice thermal conductivity enhancement at the ferroelectric phase transition in GeTe

2021 ◽  
Vol 7 (1) ◽  
Author(s):  
Đorđe Dangić ◽  
Olle Hellman ◽  
Stephen Fahy ◽  
Ivana Savić
Keyword(s):  
Phase Transition ◽  
Thermal Conductivity ◽  
Phase Transitions ◽  
Thermoelectric Materials ◽  
Ferroelectric Phase Transition ◽  
Lattice Thermal Conductivity ◽  
Ferroelectric Phase ◽  
Structural Phase ◽  
High Symmetry ◽  
Structural Phase Transitions

AbstractThe proximity to structural phase transitions in IV-VI thermoelectric materials is one of the main reasons for their large phonon anharmonicity and intrinsically low lattice thermal conductivity κ. However, the κ of GeTe increases at the ferroelectric phase transition near 700 K. Using first-principles calculations with the temperature dependent effective potential method, we show that this rise in κ is the consequence of negative thermal expansion in the rhombohedral phase and increase in the phonon lifetimes in the high-symmetry phase. Strong anharmonicity near the phase transition induces non-Lorentzian shapes of the phonon power spectra. To account for these effects, we implement a method of calculating κ based on the Green-Kubo approach and find that the Boltzmann transport equation underestimates κ near the phase transition. Our findings elucidate the influence of structural phase transitions on κ and provide guidance for design of better thermoelectric materials.

scholarly journals Thermal, ferroelastic, and structural properties near phase transitions of organic–inorganic perovskite type [NH3(CH2)3NH3]CdBr4 crystals

RSC Advances ◽  
2021 ◽  
Vol 11 (29) ◽  
pp. 17622-17629
Author(s):  
Ae Ran Lim
Keyword(s):  
Phase Transition ◽  
Phase Transitions ◽  
Thermal Behavior ◽  
Structural Properties ◽  
Structural Dynamics ◽  
Transition Temperatures ◽  
Inorganic Perovskite ◽  
Phase Transition Temperatures ◽  
Perovskite Type

We studied the thermal behavior and structural dynamics of [NH3(CH2)3NH3]CdBr4 near phase transition temperatures.

Auto-sharing parameters for transfer learning based on multi-objective optimization

2021 ◽  
pp. 1-13
Author(s):  
Hailin Liu ◽  
Fangqing Gu ◽  
Zixian Lin
Keyword(s):  
Transfer Learning ◽  
Optimization Problem ◽  
Data Sets ◽  
Multi Objective Optimization ◽  
Particle Swarm Optimizer ◽  
Real World Data ◽  
Data Set ◽  
Target Task ◽  
Main Research ◽  
Multi Objective

Transfer learning methods exploit similarities between different datasets to improve the performance of the target task by transferring knowledge from source tasks to the target task. “What to transfer” is a main research issue in transfer learning. The existing transfer learning method generally needs to acquire the shared parameters by integrating human knowledge. However, in many real applications, an understanding of which parameters can be shared is unknown beforehand. Transfer learning model is essentially a special multi-objective optimization problem. Consequently, this paper proposes a novel auto-sharing parameter technique for transfer learning based on multi-objective optimization and solves the optimization problem by using a multi-swarm particle swarm optimizer. Each task objective is simultaneously optimized by a sub-swarm. The current best particle from the sub-swarm of the target task is used to guide the search of particles of the source tasks and vice versa. The target task and source task are jointly solved by sharing the information of the best particle, which works as an inductive bias. Experiments are carried out to evaluate the proposed algorithm on several synthetic data sets and two real-world data sets of a school data set and a landmine data set, which show that the proposed algorithm is effective.

Various types of holographic metal/superconductor phase transitions with dark matter sector

2016 ◽  
Vol 26 (06) ◽  
pp. 1750046
Author(s):  
Yan Peng ◽  
Tao Chen ◽  
Guohua Liu ◽  
Pengwei Ma
Keyword(s):  
Phase Transition ◽  
Black Hole ◽  
Dark Matter ◽  
Phase Transitions ◽  
Transition Temperature ◽  
De Sitter ◽  
Holographic Superconductor ◽  
Black Hole Background ◽  
Critical Phase ◽  
Matter Sector

We generalize the holographic superconductor model with dark matter sector by including the Stückelberg mechanism in the four-dimensional anti-de Sitter (AdS) black hole background away from the probe limit. We study effects of the dark matter sector on the [Formula: see text]-wave scalar condensation and find that the dark matter sector affects the critical phase transition temperature and also the order of phase transitions. At last, we conclude that the dark matter sector brings richer physics in this general metal/superconductor system.

INDUCED CHANGES OF THERMOTROPIC PROPERTIES OF PHASE TRANSITIONS BETWEEN SMECTIC A MESOPHASE AND ISOTROPIC LIQUID: EFFECT OF SURFACES

2006 ◽  
Vol 20 (14) ◽  
pp. 821-833 ◽  
Author(s):  
ARIF NESRULLAJEV ◽  
ŞENER OKTIK
Keyword(s):  
Thin Film ◽  
Thin Films ◽  
Phase Transition ◽  
Phase Transitions ◽  
Isotropic Liquid ◽  
Smectic A ◽  
Phase Transition Temperatures ◽  
Smectic A Phase ◽  
Induced Changes ◽  
Thermotropic Properties

In this work, the effect of thin films on the thermotropic and thermo-optical properties and peculiarities of the phase transitions between the smectic A and isotropic liquid have been investigated. Peculiarities of the heterophase regions of the straight smectic A-isotropic liquid and reverse isotropic liquid-smectic A phase transitions have been studied. Change of morphologic properties of the heterophase regions, shift of the phase transition temperatures and the change of temperature widths of these heterophase regions under thin film influence have been observed.

Mössbauer Studies of Fe2+ in Iron Langbeinites and other Crystals with Langbeinite Structure

1998 ◽  
Vol 53 (1-2) ◽  
pp. 27-37 ◽  
Author(s):  
M. Windhaus ◽  
B. D. Mosel ◽  
W. Müller-Warmuth
Keyword(s):  
Phase Transition ◽  
Phase Transitions ◽  
High Temperature ◽  
Low Temperature ◽  
High Spin State ◽  
Cubic Phases ◽  
Debye Temperatures ◽  
Phase Transition Temperatures ◽  
Abrupt Changes ◽  
Temperature Dependences

Abstract 57 Fe Mössbauer spectra have been measured at various temperatures between 4.2 K and 300 K for iron langbeinites A 2 Fe 2^04)3 with A = K, NH 4 , Rb, T1 and magnesium, manganese and cadmium lang-beinites doped with Fe" + . The spectra revealed several contributions whose isomer shifts and quadru-pole splittings have been obtained by fitting program routines. For the high-temperature cubic phases two crystallographically non-equivalent iron sites have been identified, characteristic of Fe2+ in the high-spin state. Abrupt changes of the quadrupole couplings indicated phase transitions; in some cases, the spectra have also revealed several sites for Fe2+ in low temperature phases. From the temperature dependences, phase transition temperatures, crystal field splittings and Debye temperatures have been derived.

Sign in / Sign up

Export Citation Format

Share Document