-
Continuous fake media detection: adapting deepfake detectors to new generative techniques
Authors:
Francesco Tassone,
Luca Maiano,
Irene Amerini
Abstract:
Generative techniques continue to evolve at an impressively high rate, driven by the hype about these technologies. This rapid advancement severely limits the application of deepfake detectors, which, despite numerous efforts by the scientific community, struggle to achieve sufficiently robust performance against the ever-changing content. To address these limitations, in this paper, we propose an…
▽ More
Generative techniques continue to evolve at an impressively high rate, driven by the hype about these technologies. This rapid advancement severely limits the application of deepfake detectors, which, despite numerous efforts by the scientific community, struggle to achieve sufficiently robust performance against the ever-changing content. To address these limitations, in this paper, we propose an analysis of two continuous learning techniques on a Short and a Long sequence of fake media. Both sequences include a complex and heterogeneous range of deepfakes generated from GANs, computer graphics techniques, and unknown sources. Our study shows that continual learning could be important in mitigating the need for generalizability. In fact, we show that, although with some limitations, continual learning methods help to maintain good performance across the entire training sequence. For these techniques to work in a sufficiently robust way, however, it is necessary that the tasks in the sequence share similarities. In fact, according to our experiments, the order and similarity of the tasks can affect the performance of the models over time. To address this problem, we show that it is possible to group tasks based on their similarity. This small measure allows for a significant improvement even in longer sequences. This result suggests that continual techniques can be combined with the most promising detection methods, allowing them to catch up with the latest generative techniques. In addition to this, we propose an overview of how this learning approach can be integrated into a deepfake detection pipeline for continuous integration and continuous deployment (CI/CD). This allows you to keep track of different funds, such as social networks, new generative tools, or third-party datasets, and through the integration of continuous learning, allows constant maintenance of the detectors.
△ Less
Submitted 12 June, 2024;
originally announced June 2024.
-
Plasmonics in heavily-doped semiconductor nanocrystals
Authors:
Francesco Scotognella,
Giuseppe Della Valle,
Ajay Ram Srimath Kandada,
Margherita Zavelani-Rossi,
Alberto Comin,
Stefano Longhi,
Liberato Manna,
Guglielmo Lanzani,
Francesco Tassone
Abstract:
Heavily-doped semiconductor nanocrystals characterized by a tunable plasmonic band have been gaining increasing attention recently. Herein, we introduce this type of materials focusing on their structural and photo physical properties. Beside their continuous-wave plasmonic response, depicted both theoretically and experimentally, we also review recent results on their transient, ultrafast respons…
▽ More
Heavily-doped semiconductor nanocrystals characterized by a tunable plasmonic band have been gaining increasing attention recently. Herein, we introduce this type of materials focusing on their structural and photo physical properties. Beside their continuous-wave plasmonic response, depicted both theoretically and experimentally, we also review recent results on their transient, ultrafast response. This was successfully interpreted by adapting models of the ultrafast response of gold nanoparticles.
△ Less
Submitted 5 June, 2013;
originally announced June 2013.
-
Photoelasticity of sodium silicate glass from first principles
Authors:
D. Donadio,
M. Bernasconi,
F. Tassone
Abstract:
Based on density-functional perturbation theory we have computed the photoelastic tensor of a model of sodium silicate glass of composition (Na$_2$O)$_{0.25}$(SiO$_2$)$_{0.75}$ (NS3). The model (containig 84 atoms) is obtained by quenching from the melt in combined classical and Car-Parrinello molecular dynamics simulations. The calculated photoelastic coefficients are in good agreement with exp…
▽ More
Based on density-functional perturbation theory we have computed the photoelastic tensor of a model of sodium silicate glass of composition (Na$_2$O)$_{0.25}$(SiO$_2$)$_{0.75}$ (NS3). The model (containig 84 atoms) is obtained by quenching from the melt in combined classical and Car-Parrinello molecular dynamics simulations. The calculated photoelastic coefficients are in good agreement with experimental data. In particular, the calculation reproduces quantitatively the decrease of the photoelastic response induced by the insertion of Na, as measured experimentally.
The extension to NS3 of a phenomenological model developed in a previous work for pure a-SiO$_2$ indicates that the modulation upon strain of other structural parameters besides the SiOSi angles must be invoked to explain the change in the photoelstic response induced by Na.
△ Less
Submitted 9 July, 2004;
originally announced July 2004.
-
Photoelasticity of crystalline and amorphous silica from first principles
Authors:
D. Donadio,
M. Bernasconi,
F. Tassone
Abstract:
Based on density-functional perturbation theory we have computed from first principles the photoelastic tensor of few crystalline phases of silica at normal conditions and high pressure (quartz, $α$-cristobalite, $β$-cristobalite) and of models of amorphous silica (containig up to 162 atoms), obtained by quenching from the melt in combined classical and Car-Parrinello molecular dynamics simulati…
▽ More
Based on density-functional perturbation theory we have computed from first principles the photoelastic tensor of few crystalline phases of silica at normal conditions and high pressure (quartz, $α$-cristobalite, $β$-cristobalite) and of models of amorphous silica (containig up to 162 atoms), obtained by quenching from the melt in combined classical and Car-Parrinello molecular dynamics simulations. The computational framework has also been checked on the photoelastic tensor of crystalline silicon and MgO as prototypes of covalent and ionic systems. The agreement with available experimental data is good.
A phenomenological model suitable to describe the photoelastic properties of different silica polymorphs is devised by fitting on the ab-initio data.
△ Less
Submitted 5 August, 2003;
originally announced August 2003.
-
Role of bound pairs in the optical properties of highly excited semiconductors: a self consistent ladder approximation approach
Authors:
C. Piermarocchi,
F. Tassone
Abstract:
Presence of bound pairs (excitons) in a low-temperature electron-hole plasma is accounted for by including correlation between fermions at the ladder level. Using a simplified one-dimensional model with on-site Coulomb interaction, we calculate the one-particle self-energies, chemical potential, and optical response. The results are compared to those obtained in the Born approximation, which doe…
▽ More
Presence of bound pairs (excitons) in a low-temperature electron-hole plasma is accounted for by including correlation between fermions at the ladder level. Using a simplified one-dimensional model with on-site Coulomb interaction, we calculate the one-particle self-energies, chemical potential, and optical response. The results are compared to those obtained in the Born approximation, which does not account for bound pairs. In the self-consistent ladder approximation the self-energy and spectral function show a characteristic correlation peak at the exciton energy for low temperature and density. In this regime the Born approximation overestimates the chemical potential. Provided the appropriate vertex correction in the interaction with the photon is included, both ladder and Born approximations reproduce the excitonic and free pair optical absorption at low density, and the disappearance of the exciton absorption peak at larger density. However, lineshapes and energy shifts with density of the absorption and photoluminescence peaks are drastically different. In particular, the photoluminescence emission peak is much more stable in the ladder approximation. At low temperature and density a sizeable optical gain is produced in both approximations just below the excitonic peak, however this gain shows unphysical features in the Born approximation. We conclude that at low density and temperature it is fundamental to take into account the existence of bound pairs in the electron-hole plasma for the calculation of its optical and thermodynamic properties. Other approximations that fail to do so are intrinsically unphysical in this regime, and for example are not suitable to address the problem of excitonic lasing.
△ Less
Submitted 25 September, 2000;
originally announced September 2000.
-
Evidence for Stimulated Scattering of Excitons into Microcavity Polaritons
Authors:
F. Tassone,
R. Huang,
Y. Yamamoto
Abstract:
The wavefunction of a collection of identical quantum particles of integer spin (bosons) is even under exchange of the coordinates of any two of them. This symmetrization rule of quantum mechanics leads to stimulation of the scattering rate into final states whose population is larger than one. For massless photons, it is routinely observed in the laser. For massive bosons, most previous efforts…
▽ More
The wavefunction of a collection of identical quantum particles of integer spin (bosons) is even under exchange of the coordinates of any two of them. This symmetrization rule of quantum mechanics leads to stimulation of the scattering rate into final states whose population is larger than one. For massless photons, it is routinely observed in the laser. For massive bosons, most previous efforts concentrated in realizing the largely populated equilibrium Bose-Einstein condensate, and observing stimulated emission into it. Here we show that we do not need to produce such a condensate to observe final state stimulation. Instead, we take advantage of the highly non-equilibrium exciton reservoir and of the large polariton populations produced by external laser action on a semiconductor microcavity, and measure final state stimulation of the scattering rate of these excitons into the polaritons.
△ Less
Submitted 14 August, 1998;
originally announced August 1998.
-
Electron-hole correlation effects in the emission of light from quantum wires
Authors:
Francesco Tassone,
Carlo Piermarocchi
Abstract:
We present a self-consistent treatment of the electron-hole correlations in optically excited quantum wires within the ladder approximation, and using a contact potential interaction. The limitations of the ladder approximation to the excitonic low-density region are largely overcome by the introduction of higher order correlations through self consistency. We show relevance of these correlation…
▽ More
We present a self-consistent treatment of the electron-hole correlations in optically excited quantum wires within the ladder approximation, and using a contact potential interaction. The limitations of the ladder approximation to the excitonic low-density region are largely overcome by the introduction of higher order correlations through self consistency. We show relevance of these correlations in the low-temperature emission, even for high density relevant in lasing, when large gain replaces excitonic absorption.
△ Less
Submitted 14 August, 1998;
originally announced August 1998.
-
Acceleration Schemes for Ab-Initio Molecular Dynamics and Electronic Structure Calculations
Authors:
Francesco Tassone,
Francesco Mauri,
Roberto Car
Abstract:
We study the convergence and the stability of fictitious dynamical methods for electrons. First, we show that a particular damped second-order dynamics has a much faster rate of convergence to the ground-state than first-order steepest descent algorithms while retaining their numerical cost per time step. Our damped dynamics has efficiency comparable to that of conjugate gradient methods in typi…
▽ More
We study the convergence and the stability of fictitious dynamical methods for electrons. First, we show that a particular damped second-order dynamics has a much faster rate of convergence to the ground-state than first-order steepest descent algorithms while retaining their numerical cost per time step. Our damped dynamics has efficiency comparable to that of conjugate gradient methods in typical electronic minimization problems. Then, we analyse the factors that limit the size of the integration time step in approaches based on plane-wave expansions. The maximum allowed time step is dictated by the highest frequency components of the fictitious electronic dynamics. These can result either from the large wavevector components of the kinetic energy or from the small wavevector components of the Coulomb potential giving rise to the so called {\it charge sloshing} problem. We show how to eliminate large wavevector instabilities by adopting a preconditioning scheme that is implemented here for the first-time in the context of Car-Parrinello ab-initio molecular dynamics simulations of the ionic motion. We also show how to solve the charge-sloshing problem when this is present. We substantiate our theoretical analysis with numerical tests on a number of different silicon and carbon systems having both insulating and metallic character.
△ Less
Submitted 23 August, 1994;
originally announced August 1994.
