Search

Training machine learning models on data from multiple entities without direct data sharing can unlock applications otherwise hindered by business, legal, or ethical constraints. In this work, we design and implement new privacy-preserving machine learning protocols for logistic regression and neural network models. We adopt a two-server model where data owners secret-share their data between two servers that train and evaluate the model on the joint data. A significant source of inefficiency and inaccuracy in existing methods arises from using Yao's garbled circuits to compute non-linear activation functions. We propose new methods for computing non-linear functions based on secret-shared lookup tables, offering both computational efficiency and improved accuracy. Beyond introducing leakage-free techniques, we initiate the exploration of relaxed security measures for privacy-preserving machine learning. Instead of claiming that the servers gain no knowledge during the computation, we contend that while some information is revealed about access patterns to lookup tables, it maintains epsilon-dX-privacy. Leveraging this relaxation significantly reduces the computational resources needed for training. We present new cryptographic protocols tailored to this relaxed security paradigm and define and analyze the leakage. Our evaluations show that our logistic regression protocol is up to 9x faster, and the neural network training is up to 688x faster than SecureML. Notably, our neural network achieves an accuracy of 96.6% on MNIST in 15 epochs, outperforming prior benchmarks that capped at 93.4% using the same architecture., Comment: Accepted at Privacy Enhancing Technologies Symposium (PETS) 2024

We have investigated the effects of high-energy electronirradiationon the oxidation of copper nanoparticles in environmental scanningtransmission electron microscopy (ESTEM). The hemispherically shapedparticles were oxidized in 3 mbar of O-2 in a temperaturerange 100-200 & DEG;C. The evolution of the particles was recordedwith sub-nanometer spatial resolution in situ in ESTEM. The oxidationencompasses the formation of outer and inner oxide shells on the nanoparticles,arising from the concurrent diffusion of copper and oxygen out ofand into the nanoparticles, respectively. Our results reveal thatthe electron beam actively influences the reaction and overall acceleratesthe oxidation of the nanoparticles when compared to particles oxidizedwithout exposure to the electron beam. However, the extent of thiselectron beam-assisted acceleration of oxidation diminishes at highertemperatures. Moreover, we observe that while oxidation through theoutward diffusion of Cu+ cations is enhanced, the electronbeam appears to hinder oxidation through the inward diffusion of O2- anions. Our results suggest that the impact of thehigh-energy electrons in ESTEM oxidation of Cu nanoparticles is mostlyrelated to kinetic energy transfer, charging, and ionization of thegas environment, and the beam can both enhance and suppress reactionrates., Funding Agencies|VILLUM FONDEN [36155]; Knut and Alice Wallenberg Foundation [2015.0055]; Swedish Research Council (VR) [2018-00329]

We have investigated the effects of high-energy electronirradiationon the oxidation of copper nanoparticles in environmental scanningtransmission electron microscopy (ESTEM). The hemispherically shapedparticles were oxidized in 3 mbar of O-2 in a temperaturerange 100-200 & DEG;C. The evolution of the particles was recordedwith sub-nanometer spatial resolution in situ in ESTEM. The oxidationencompasses the formation of outer and inner oxide shells on the nanoparticles,arising from the concurrent diffusion of copper and oxygen out ofand into the nanoparticles, respectively. Our results reveal thatthe electron beam actively influences the reaction and overall acceleratesthe oxidation of the nanoparticles when compared to particles oxidizedwithout exposure to the electron beam. However, the extent of thiselectron beam-assisted acceleration of oxidation diminishes at highertemperatures. Moreover, we observe that while oxidation through theoutward diffusion of Cu+ cations is enhanced, the electronbeam appears to hinder oxidation through the inward diffusion of O2- anions. Our results suggest that the impact of thehigh-energy electrons in ESTEM oxidation of Cu nanoparticles is mostlyrelated to kinetic energy transfer, charging, and ionization of thegas environment, and the beam can both enhance and suppress reactionrates., Funding Agencies|VILLUM FONDEN [36155]; Knut and Alice Wallenberg Foundation [2015.0055]; Swedish Research Council (VR) [2018-00329]

We have investigated the effects of high-energy electronirradiationon the oxidation of copper nanoparticles in environmental scanningtransmission electron microscopy (ESTEM). The hemispherically shapedparticles were oxidized in 3 mbar of O-2 in a temperaturerange 100-200 & DEG;C. The evolution of the particles was recordedwith sub-nanometer spatial resolution in situ in ESTEM. The oxidationencompasses the formation of outer and inner oxide shells on the nanoparticles,arising from the concurrent diffusion of copper and oxygen out ofand into the nanoparticles, respectively. Our results reveal thatthe electron beam actively influences the reaction and overall acceleratesthe oxidation of the nanoparticles when compared to particles oxidizedwithout exposure to the electron beam. However, the extent of thiselectron beam-assisted acceleration of oxidation diminishes at highertemperatures. Moreover, we observe that while oxidation through theoutward diffusion of Cu+ cations is enhanced, the electronbeam appears to hinder oxidation through the inward diffusion of O2- anions. Our results suggest that the impact of thehigh-energy electrons in ESTEM oxidation of Cu nanoparticles is mostlyrelated to kinetic energy transfer, charging, and ionization of thegas environment, and the beam can both enhance and suppress reactionrates., Funding Agencies|VILLUM FONDEN [36155]; Knut and Alice Wallenberg Foundation [2015.0055]; Swedish Research Council (VR) [2018-00329]

We have investigated the effects of high-energy electron irradiation on the oxidation of copper nanoparticles in environmental scanning transmission electron microscopy (ESTEM). The hemispherically shaped particles were oxidized in 3 mbar of O2 in a temperature range 100–200 °C. The evolution of the particles was recorded with sub-nanometer spatial resolution in situ in ESTEM. The oxidation encompasses the formation of outer and inner oxide shells on the nanoparticles, arising from the concurrent diffusion of copper and oxygen out of and into the nanoparticles, respectively. Our results reveal that the electron beam actively influences the reaction and overall accelerates the oxidation of the nanoparticles when compared to particles oxidized without exposure to the electron beam. However, the extent of this electron beam-assisted acceleration of oxidation diminishes at higher temperatures. Moreover, we observe that while oxidation through the outward diffusion of Cu+ cations is enhanced, the electron beam appears to hinder oxidation through the inward diffusion of O2– anions. Our results suggest that the impact of the high-energy electrons in ESTEM oxidation of Cu nanoparticles is mostly related to kinetic energy transfer, charging, and ionization of the gas environment, and the beam can both enhance and suppress reaction rates.

We have investigated the effects of high-energy electronirradiationon the oxidation of copper nanoparticles in environmental scanningtransmission electron microscopy (ESTEM). The hemispherically shapedparticles were oxidized in 3 mbar of O-2 in a temperaturerange 100-200 & DEG;C. The evolution of the particles was recordedwith sub-nanometer spatial resolution in situ in ESTEM. The oxidationencompasses the formation of outer and inner oxide shells on the nanoparticles,arising from the concurrent diffusion of copper and oxygen out ofand into the nanoparticles, respectively. Our results reveal thatthe electron beam actively influences the reaction and overall acceleratesthe oxidation of the nanoparticles when compared to particles oxidizedwithout exposure to the electron beam. However, the extent of thiselectron beam-assisted acceleration of oxidation diminishes at highertemperatures. Moreover, we observe that while oxidation through theoutward diffusion of Cu+ cations is enhanced, the electronbeam appears to hinder oxidation through the inward diffusion of O2- anions. Our results suggest that the impact of thehigh-energy electrons in ESTEM oxidation of Cu nanoparticles is mostlyrelated to kinetic energy transfer, charging, and ionization of thegas environment, and the beam can both enhance and suppress reactionrates., Funding Agencies|VILLUM FONDEN [36155]; Knut and Alice Wallenberg Foundation [2015.0055]; Swedish Research Council (VR) [2018-00329]

We have investigated the effects of high-energy electronirradiationon the oxidation of copper nanoparticles in environmental scanningtransmission electron microscopy (ESTEM). The hemispherically shapedparticles were oxidized in 3 mbar of O-2 in a temperaturerange 100-200 & DEG;C. The evolution of the particles was recordedwith sub-nanometer spatial resolution in situ in ESTEM. The oxidationencompasses the formation of outer and inner oxide shells on the nanoparticles,arising from the concurrent diffusion of copper and oxygen out ofand into the nanoparticles, respectively. Our results reveal thatthe electron beam actively influences the reaction and overall acceleratesthe oxidation of the nanoparticles when compared to particles oxidizedwithout exposure to the electron beam. However, the extent of thiselectron beam-assisted acceleration of oxidation diminishes at highertemperatures. Moreover, we observe that while oxidation through theoutward diffusion of Cu+ cations is enhanced, the electronbeam appears to hinder oxidation through the inward diffusion of O2- anions. Our results suggest that the impact of thehigh-energy electrons in ESTEM oxidation of Cu nanoparticles is mostlyrelated to kinetic energy transfer, charging, and ionization of thegas environment, and the beam can both enhance and suppress reactionrates., Funding Agencies|VILLUM FONDEN [36155]; Knut and Alice Wallenberg Foundation [2015.0055]; Swedish Research Council (VR) [2018-00329]

In today's world, social networks have become one of the primary sources for creation and propagation of news. Social news aggregators are one of the actors in this area in which users post news items and use positive or negative votes to indicate their preference toward a news item. News items will be ordered and displayed according to their aggregated votes. This approach suffers from several problems raging from being prone to the dominance of the majority to difficulty in discerning between correct and fake news, and lack of incentive for honest behaviors. In this paper, we propose a graph-based news aggregator in which instead of voting on the news items, users submit their votes on the relations between pairs of news items. More precisely, if a user believes two news items support each other, he will submit a positive vote on the link between the two items, and if he believes that two news items undermine each other, he will submit a negative vote on the corresponding link. This approach has mainly two desirable features: (1) mitigating the effect of personal preferences on voting, (2) connection of new items to endorsing and disputing evidence. This approach helps the newsreaders to understand different aspects of a news item better. We also introduce an incentive layer that uses blockchain as a distributed transparent manager to encourages users to behave honestly and abstain from adversary behaviors. The incentive layer takes into account that users can have different viewpoints toward news, enabling users from a wide range of viewpoints to contribute to the network and benefit from its rewards. In addition, we introduce a protocol that enables us to prove fraud in computations of the incentive layer model on the blockchain. Ultimately, we will analyze the fraud proof protocol and examine our incentive layer on a wide range of synthesized datasets., Comment: 23 page, 8 figures, Abstract abridged due to arXiv limits