Mobile authentication of copy detection patterns.

In the recent years, the copy detection patterns (CDP) attracted a lot of attention as a link between the physical and digital worlds, which is of great interest for the internet of things and brand protection applications. However, the security of CDP in terms of their reproducibility by unauthorized parties or clonability remains largely unexplored. In this respect, this paper addresses a problem of anti-counterfeiting of physical objects and aims at investigating the authentication aspects and the resistances to illegal copying of the modern CDP from machine learning perspectives. A special attention is paid to a reliable authentication under the real-life verification conditions when the codes are printed on an industrial printer and enrolled via modern mobile phones under regular light conditions. The theoretical and empirical investigation of authentication aspects of CDP is performed with respect to four types of copy fakes from the point of view of (i) multi-class supervised classification as a baseline approach and (ii) one-class classification as a real-life application case. The obtained results show that the modern machine-learning approaches and the technical capacities of modern mobile phones allow to reliably authenticate CDP on end-user mobile phones under the considered classes of fakes.

Multivariate Time Series Information Bottleneck.

Time series (TS) and multiple time series (MTS) predictions have historically paved the way for distinct families of deep learning models. The temporal dimension, distinguished by its evolutionary sequential aspect, is usually modeled by decomposition into the trio of "trend, seasonality, noise", by attempts to copy the functioning of human synapses, and more recently, by transformer models with self-attention on the temporal dimension. These models may find applications in finance and e-commerce, where any increase in performance of less than 1% has large monetary repercussions, they also have potential applications in natural language processing (NLP), medicine, and physics. To the best of our knowledge, the information bottleneck (IB) framework has not received significant attention in the context of TS or MTS analyses. One can demonstrate that a compression of the temporal dimension is key in the context of MTS. We propose a new approach with partial convolution, where a time sequence is encoded into a two-dimensional representation resembling images. Accordingly, we use the recent advances made in image extension to predict an unseen part of an image from a given one. We show that our model compares well with traditional TS models, has information-theoretical foundations, and can be easily extended to more dimensions than only time and space. An evaluation of our multiple time series-information bottleneck (MTS-IB) model proves its efficiency in electricity production, road traffic, and astronomical data representing solar activity, as recorded by NASA's interface region imaging spectrograph (IRIS) satellite.

Variational Information Bottleneck for Semi-Supervised Classification.

In this paper, we consider an information bottleneck (IB) framework for semi-supervised classification with several families of priors on latent space representation. We apply a variational decomposition of mutual information terms of IB. Using this decomposition we perform an analysis of several regularizers and practically demonstrate an impact of different components of variational model on the classification accuracy. We propose a new formulation of semi-supervised IB with hand crafted and learnable priors and link it to the previous methods such as semi-supervised versions of VAE (M1 + M2), AAE, CatGAN, etc. We show that the resulting model allows better understand the role of various previously proposed regularizers in semi-supervised classification task in the light of IB framework. The proposed IB semi-supervised model with hand-crafted and learnable priors is experimentally validated on MNIST under different amount of labeled data.

Information Bottleneck Classification in Extremely Distributed Systems.

We present a new decentralized classification system based on a distributed architecture. This system consists of distributed nodes, each possessing their own datasets and computing modules, along with a centralized server, which provides probes to classification and aggregates the responses of nodes for a final decision. Each node, with access to its own training dataset of a given class, is trained based on an auto-encoder system consisting of a fixed data-independent encoder, a pre-trained quantizer and a class-dependent decoder. Hence, these auto-encoders are highly dependent on the class probability distribution for which the reconstruction distortion is minimized. Alternatively, when an encoding-quantizing-decoding node observes data from different distributions, unseen at training, there is a mismatch, and such a decoding is not optimal, leading to a significant increase of the reconstruction distortion. The final classification is performed at the centralized classifier that votes for the class with the minimum reconstruction distortion. In addition to the system applicability for applications facing big-data communication problems and or requiring private classification, the above distributed scheme creates a theoretical bridge to the information bottleneck principle. The proposed system demonstrates a very promising performance on basic datasets such as MNIST and FasionMNIST.

Machine learning through cryptographic glasses: combating adversarial attacks by key-based diversified aggregation.

In recent years, classification techniques based on deep neural networks (DNN) were widely used in many fields such as computer vision, natural language processing, and self-driving cars. However, the vulnerability of the DNN-based classification systems to adversarial attacks questions their usage in many critical applications. Therefore, the development of robust DNN-based classifiers is a critical point for the future deployment of these methods. Not less important issue is understanding of the mechanisms behind this vulnerability. Additionally, it is not completely clear how to link machine learning with cryptography to create an information advantage of the defender over the attacker. In this paper, we propose a key-based diversified aggregation (KDA) mechanism as a defense strategy in a gray- and black-box scenario. KDA assumes that the attacker (i) knows the architecture of classifier and the used defense strategy, (ii) has an access to the training data set, but (iii) does not know a secret key and does not have access to the internal states of the system. The robustness of the system is achieved by a specially designed key-based randomization. The proposed randomization prevents the gradients' back propagation and restricts the attacker to create a "bypass" system. The randomization is performed simultaneously in several channels. Each channel introduces its own randomization in a special transform domain. The sharing of a secret key between the training and test stages creates an information advantage to the defender. Finally, the aggregation of soft outputs from each channel stabilizes the results and increases the reliability of the final score. The performed experimental evaluation demonstrates a high robustness and universality of the KDA against state-of-the-art gradient-based gray-box transferability attacks and the non-gradient-based black-box attacks (The results reported in this paper have been partially presented in CVPR 2019 (Taran et al., Defending against adversarial attacks by randomized diversification, 2019) & ICIP 2019 (Taran et al., Robustification of deep net classifiers by key-based diversified aggregation with pre-filtering, 2019)).

Electrostatic Complementarity Drives Amyloid/Nucleic Acid Co-Assembly.

Proteinaceous plaques associated with neurodegenerative diseases contain many biopolymers including the polyanions glycosaminoglycans and nucleic acids. Polyanion-induced amyloid fibrillation has been implicated in disease etiology, but structural models for amyloid/nucleic acid co-assemblies remain limited. Here we constrain nucleic acid/peptide interactions with model peptides that exploit electrostatic complementarity and define a novel amyloid/nucleic acid co-assembly. The structure provides a model for nucleic acid/amyloid co-assembly as well as insight into the energetic determinants involved in templating amyloid assembly.

Small molecule reaction networks that model the ROS dynamics of the rhizosphere.

Spontaneous reactions between plant and bacterial redox active metabolites can result in reaction-diffusion networks that regulate redox gradients and ROS concentrations. Our model system mimics known biological processes observed in plants, including the oxidative burst, travelling waves, and chemical pattern formation. Similar non-enzymatic reactions between natural products may play a role in plant-bacteria interactions, including biofilm and microbiome regulation, and be useful for the development of narrow range antibiotics.

Correction: Achieving biopolymer synergy in systems chemistry.

Correction for 'Achieving biopolymer synergy in systems chemistry' by Yushi Bai et al., Chem. Soc. Rev., 2018, DOI: 10.1039/c8cs00174j.

Achieving biopolymer synergy in systems chemistry.

Synthetic and materials chemistry initiatives have enabled the translation of the macromolecular functions of biology into synthetic frameworks. These explorations into alternative chemistries of life attempt to capture the versatile functionality and adaptability of biopolymers in new orthogonal scaffolds. Information storage and transfer, however, so beautifully represented in the central dogma of biology, require multiple components functioning synergistically. Over a single decade, the emerging field of systems chemistry has begun to catalyze the construction of mutualistic biopolymer networks, and this review begins with the foundational small-molecule-based dynamic chemical networks and peptide amyloid-based dynamic physical networks on which this effort builds. The approach both contextualizes the versatile approaches that have been developed to enrich chemical information in synthetic networks and highlights the properties of amyloids as potential alternative genetic elements. The successful integration of both chemical and physical networks through ß-sheet assisted replication processes further informs the synergistic potential of these networks. Inspired by the cooperative synergies of nucleic acids and proteins in biology, synthetic nucleic-acid-peptide chimeras are now being explored to extend their informational content. With our growing range of synthetic capabilities, structural analyses, and simulation technologies, this foundation is radically extending the structural space that might cross the Darwinian threshold for the origins of life as well as creating an array of alternative systems capable of achieving the progressive growth of novel informational materials.

Expanding the informational chemistries of life: peptide/RNA networks.

The RNA world hypothesis simplifies the complex biopolymer networks underlining the informational and metabolic needs of living systems to a single biopolymer scaffold. This simplification requires abiotic reaction cascades for the construction of RNA, and this chemistry remains the subject of active research. Here, we explore a complementary approach involving the design of dynamic peptide networks capable of amplifying encoded chemical information and setting the stage for mutualistic associations with RNA. Peptide conformational networks are known to be capable of evolution in disease states and of co-opting metal ions, aromatic heterocycles and lipids to extend their emergent behaviours. The coexistence and association of dynamic peptide and RNA networks appear to have driven the emergence of higher-order informational systems in biology that are not available to either scaffold independently, and such mutualistic interdependence poses critical questions regarding the search for life across our Solar System and beyond.This article is part of the themed issue 'Reconceptualizing the origins of life'.

Redox-mediated quorum sensing in plants.

The rhizosphere, the narrow zone of soil around plant roots, is a complex network of interactions between plants, bacteria, and a variety of other organisms. The absolute dependence on host-derived signals, or xenognosins, to regulate critical developmental checkpoints for host commitment in the obligate parasitic plants provides a window into the rhizosphere's chemical dynamics. These sessile intruders use H2O2 in a process known as semagenesis to chemically modify the mature root surfaces of proximal host plants and generate p-benzoquinones (BQs). The resulting redox-active signaling network regulates the spatial and temporal commitments necessary for host attachment. Recent evidence from non-parasites, including Arabidopsis thaliana, establishes that reactive oxygen species (ROS) production regulates similar redox circuits related to root recognition, broadening xenognosins' role beyond the parasites. Here we compare responses to the xenognosin dimethoxybenzoquinone (DMBQ) between the parasitic plant Striga asiatica and the non-parasitic A. thaliana. Exposure to DMBQ simulates the proximity of a mature root surface, stimulating an increase in cytoplasmic Ca2+ concentration in both plants, but leads to remarkably different phenotypic responses in the parasite and non-parasite. In S. asiatica, DMBQ induces development of the host attachment organ, the haustorium, and decreases ROS production at the root tip, while in A. thaliana, ROS production increases and further growth of the root tip is arrested. Obstruction of Ca2+ channels and the addition of antioxidants both lead to a decrease in the DMBQ response in both parasitic and non-parasitic plants. These results are consistent with Ca2+ regulating the activity of NADPH oxidases, which in turn sustain the autocatalytic production of ROS via an external quinone/hydroquinone redox cycle. Mechanistically, this chemistry is similar to black and white photography with the emerging dynamic reaction-diffusion network laying the foundation for the precise temporal and spatial control underlying rhizosphere architecture.

Electron Transfer between Electrically Conductive Minerals and Quinones.

Long-distance electron transfer in marine environments couples physically separated redox half-reactions, impacting biogeochemical cycles of iron, sulfur and carbon. Bacterial bio-electrochemical systems that facilitate electron transfer via conductive filaments or across man-made electrodes are well-known, but the impact of abiotic currents across naturally occurring conductive and semiconductive minerals is poorly understood. In this paper I use cyclic voltammetry to explore electron transfer between electrodes made of common iron minerals (magnetite, hematite, pyrite, pyrrhotite, mackinawite, and greigite), and hydroquinones-a class of organic molecules found in carbon-rich sediments. Of all tested minerals, only pyrite and magnetite showed an increase in electric current in the presence of organic molecules, with pyrite showing excellent electrocatalytic performance. Pyrite electrodes performed better than commercially available glassy carbon electrodes and showed higher peak currents, lower overpotential values and a smaller separation between oxidation and reduction peaks for each tested quinone. Hydroquinone oxidation on pyrite surfaces was reversible, diffusion controlled, and stable over a large number of potential cycles. Given the ubiquity of both pyrite and quinones, abiotic electron transfer between minerals and organic molecules is likely widespread in Nature and may contribute to several different phenomena, including anaerobic respiration of a wide variety of microorganisms in temporally anoxic zones or in the proximity of hydrothermal vent chimneys, as well as quinone cycling and the propagation of anoxic zones in organic rich waters. Finally, interactions between pyrite and quinones make use of electrochemical gradients that have been suggested as an important source of energy for the origins of life on Earth. Ubiquinones and iron sulfide clusters are common redox cofactors found in electron transport chains across all domains of life and interactions between quinones and pyrite might have been an early analog of these ubiquitous systems.

Measuring binding of protein to gel-bound ligands using magnetic levitation.

This paper describes the use of magnetic levitation (MagLev) to measure the association of proteins and ligands. The method starts with diamagnetic gel beads that are functionalized covalently with small molecules (putative ligands). Binding of protein to the ligands within the bead causes a change in the density of the bead. When these beads are suspended in a paramagnetic aqueous buffer and placed between the poles of two NbFeB magnets with like poles facing, the changes in the density of the bead on binding of protein result in changes in the levitation height of the bead that can be used to quantify the amount of protein bound. This paper uses a reaction-diffusion model to examine the physical principles that determine the values of rate and equilibrium constants measured by this system, using the well-defined model system of carbonic anhydrase and aryl sulfonamides. By tuning the experimental protocol, the method is capable of quantifying either the concentration of protein in a solution, or the binding affinities of a protein to several resin-bound small molecules simultaneously. Since this method requires no electricity and only a single piece of inexpensive equipment, it may find use in situations where portability and low cost are important, such as in bioanalysis in resource-limited settings, point-of-care diagnosis, veterinary medicine, and plant pathology. It still has several practical disadvantages. Most notably, the method requires relatively long assay times and cannot be applied to large proteins (>70 kDa), including antibodies. The design and synthesis of beads with improved characteristics (e.g., larger pore size) has the potential to resolve these problems.

Mechanism of general acid-base catalysis in transesterification of an RNA model phosphodiester studied with strongly basic catalysts.

Using 80% vol aqueous DMSO as the reaction medium for transesterification of an RNA model substrate 2-hydroxypropyl 4-nitrophenyl phosphate allows one to observe catalysis in buffer mixtures composed of highly basic components such as guanidines, amidines or alkylamines, which provide up to 10(3)-fold accelerations over the background reaction in the 0.01-0.1 M concentration range. The rate law k(obs) = k(1)[B] + k(2)[B][BH(+)] was established indicating contributions from both simple general base catalysis and the reaction involving concerted action of neutral (B) and protonated (BH(+)) forms of the buffer. The catalytic efficiency of guanidinium and amidinium cations is 10 times larger than that of more acidic ammonium cations. Rate constants k(1) and k(2) obey the Brønsted equations with the slopes 0.77 and 0.69 respectively. Proton inventory for k(2) (B = guanidine) in D(2)O/H(2)O mixtures gives two fractionation factors phi(1) = 0.48 and phi(2) = 1.26 for normal and inverse isotope effects respectively. The former results from the proton transfer to B and the latter from the binding of guanidinium cation to the phosphate group as follows from observation of an inverse solvent isotope effect for the binding of guanidinium and amidinium cations to a phosphodiester anion. The results of kinetic studies together with analysis of transition state stabilization free energies for guanidinium and amidinium cations show that the protonated buffer component acts via electrostatic transition state stabilization rather than proton transfer, which may be possible for a guanidinium assisted hydroxide catalyzed reaction.

Rapid hydrolysis of model phosphate diesters by alkaline-earth cations in aqueous DMSO: speciation and kinetics.

Kinetics of the cleavage of two phosphate diesters, bis(4-nitrophenyl) phosphate and 2-hydroxypropyl 4-nitrophenyl phosphate and a triester, 4-nitrophenyl diphenyl phosphate, in the presence of Mg(II), Ca(II) and Sr(II) were studied in 90% vol. DMSO at 37 degrees C. The alkaline hydrolysis of the triester was inhibited by all cations, but with both phosphodiesters strong catalytic effects were observed. Potentiometric titrations of metal perchlorates by Bu4N(OH) revealed formation of M2(OH)3+, M(OH)+, M(OH)2 and M2(OH)5- species. Rate constants for phosphodiester cleavage by individual species were obtained from analysis of rate-concentration profiles. Observed first-order rate constants in the presence of 1-2 mM Mg(II) or Ca(II) in neutral and weakly basic solutions were 10(8)-10(11) times higher than those for background hydrolysis at the same pH while in water additions of up to 50 mM metal produced <100-fold accelerations. Possible structures of DMSO solvated catalyst-substrate complexes were modeled by DFT calculations with Mg(II). The increased catalytic activity in 90% DMSO is attributed to stronger association of hydroxide ions and anionic phosphodiesters with metal ions and to preferable solvation of cations by DMSO, which creates favorable for reaction anhydrous microenvironment in the coordination sphere of the catalyst.

Solvent effects and alkali metal ion catalysis in phosphodiester hydrolysis.

The kinetics of the alkaline hydrolysis of bis(p-nitrophenyl) phosphate (BNPP) have been studied in aqueous DMSO, dioxane, and MeCN. In all solvent mixtures the reaction rate steadily decreases to half of its value in pure water in the range of 0-70 vol % of organic cosolvent and sharply increases in mixtures with lower water content. Correlations based on different scales of solvent empirical parameters failed to describe the solvent effect in this system, but it can be satisfactorily treated in terms of a simplified stepwise solvent-exchange model. Alkali metal ions catalyze the BNPP hydrolysis but do not affect the rate of hydrolysis of neutral phosphotriester p-nitrophenyl diphenyl phosphate in DMSO-rich mixtures. The catalytic activity decreases in the order Li+ > Na+ > K+ > Rb+ > Cs+. For all cations except Na+, the reaction rate is first-order in metal ion. With Na+, both first- and second-order kinetics in metal ions are observed. Binding constants of cations to the dianionic transition state of BNPP alkaline hydrolysis are of the same order of magnitude and show a similar trend as their binding constants to p-nitrophenyl phosphate dianion employed as a transition-state model. The appearance of alkali metal ion catalysis in a medium, which solvates metal ions stronger than water, is attributed to the increased affinity of cations to dianions, which undergo a strong destabilization in the presence of an aprotic dipolar cosolvent.