FORTH Institute of Astrophysics

Widespread adoption of artificial intelligence (AI) technologies is substantially affecting the human condition in ways that are not yet well understood. Negative unintended consequences abound including the perpetuation and exacerbation of societal inequalities and divisions via algorithmic decision making. We present six grand challenges for the scientific community to create AI technologies that are human-centered, that is, ethical, fair, and enhance the human condition. These grand challenges are the result of an international collaboration across academia, industry and government and represent the consensus views of a group of 26 experts in the field of human-centered artificial intelligence (HCAI). In essence, these challenges advocate for a human-centered approach to AI that (1) is centered in human well-being, (2) is designed responsibly, (3) respects privacy, (4) follows human-centered design principles, (5) is subject to appropriate governance and oversight, and (6) interacts with individuals while respecting human’s cognitive capacities. We hope that these challenges and their associated research directions serve as a call for action to conduct research and development in AI that serves as a force multiplier towards more fair, equitable and sustainable societies.

Antivirus software are the most popular tools for detecting and stopping malicious or unwanted files. However, the performance requirements of traditional host-based antivirus make their wide adoption to mobile, embedded, and hand-held devices questionable. Their computational- and memory-intensive characteristics, which are needed to cope with the evolved and sophisticated malware, makes their deployment to mobile processors a hard task. Moreover, their increasing complexity may result in vulnerabilities that can be exploited by malware.

A foundational issue underlying many overlay network applications ranging from routing to peer-to-peer file sharing is that of connectivity management, i.e., folding new arrivals into an existing overlay, and re-wiring to cope with changing network conditions. Previous work has considered the problem from two perspectives: devising practical heuristics for specific applications designed to work well in real deployments, and providing abstractions for the underlying problem that are analytically tractable, especially via game-theoretic analysis. In this paper, we unify these two thrusts by using insights gleaned from novel, realistic theoretic models in the design of EGOIST -- a distributed overlay routing system that we implemented, deployed, and evaluated on PlanetLab. Using extensive measurements of paths between nodes, we demonstrate that EGOIST'S neighbor selection primitives significantly outperform existing heuristics on a variety of performance metrics, including delay, available bandwidth, and node utilization. Moreover, we demonstrate that EGOIST is competitive with an optimal, but unscalable full-mesh approach, remains highly effective under significant churn, is robust to cheating, and incurs minimal overhead. Finally, we use a multiplayer peer-to-peer game to demonstrate the value of EGOIST to end-user applications.

We propose a novel hybrid human 3D body pose estimation method that uses RGBD input. The method relies on a deep neural network to get an initial 2D body pose. Using depth information from the sensor, a set of 2D landmarks on the body are transformed in 3D. Then, a multiple hypothesis tracker uses the obtained 2D and 3D body landmarks to estimate the 3D body pose. In order to safeguard from observation errors, each human pose hypothesis considered by the tracker is constructed using a gradient descent optimization scheme that is applied to a subset of the body landmarks. Landmark selection is driven by a set of geometric constraints and temporal continuity criteria. The resulting 3D poses are evaluated by an objective function that calculates densely the discrepancy between the 3D structure of the rendered 3D human body model and the actual depth observed by the sensor. The quantitative experiments show the advantages of the proposed method over a baseline that directly uses all landmark observations for the optimization, as well as over other recent 3D human pose estimation approaches.

Black-hole transients exhibit a correlation between the time lag of hard photons with respect to softer ones and the photon index of the hard X-ray power law. The correlation is not very tight and therefore it is necessary to examine it source by source. The objective of the present work is to investigate in detail the time-lag -- photon-index correlation in GX 339-4. We have obtained RXTE energy spectra and light curves and have computed the photon index and the time lag of the $9 - 15$ keV photons with respect to the $2 - 6$ keV ones. The observations cover the first stages of the hard state, the pure hard state, and the hard-intermediate state. At low $Γ$, the correlation is positive and it becomes negative at large $Γ$. By assuming that the hard X-ray power law index $Γ$ is produced by inverse Compton scattering of soft disk photons in the jet, we have reproduced the entire correlation by varying two parameters in the jet: the radius of the jet at its base $R_0$ and the Thomson optical depth along the jet $τ_\parallel$. We have found that, as the luminosity of the source increases, $R_0$ initially increases and then decreases. This behavior is expected in the context of the Cosmic Battery. As a further test of our model, we predict the break frequency in the radio spectrum as a function of the photon index during the rising part of an outburst.

Internet routes - controlled by the Border Gateway Protocol (BGP) - carry our communication and our commerce, yet many aspects of routing are opaque to even network operators, and BGP is known to contribute to performance, reliability, and security problems. The research and operations communities have developed a set of tools and data sources for understanding and experimenting with BGP, and on February 2016 we organized the first BGP Hackathon, themed around live measurement and monitoring of Internet routing. The Hackathon included students, researchers, operators, providers, policymakers, and funding agencies, working together on projects to measure, visualize, and improve routing or the tools we use to study routing. This report describes the tools used at the Hackathon and presents an overview of the projects. The Hackathon was a success, and we look forward to future iterations.

INTEGRAL monitoring of the Galactic Plane is revealing a growing number of recurrent X-ray transients, characterised by short outbursts with very fast rise times (~ tens of minutes) and typical durations of a few hours. A substantial fraction of these sources are associated with OB supergiants and hence define a new class of massive X-ray binaries, which we call Supergiant Fast X-ray Transients. Characterisation of the astrophysical parameters of their counterparts is underway. So far, we have found a number of late O and early B supergiants of different luminosities at a large range of distances. Nothing in their optical properties sets them apart from classical Supergiant X-ray Binaries. On the other hand, there is now rather concluding evidence that persistent supergiant X-ray binaries also show fast outbursts. This suggests a continuum of behaviours between typical persistent supergiant systems and purely transient systems, but offers very little information about the physical causes of the outbursts.

Security solutions for cloud applications usually exploit security tools as is by utilising their default configuration. On one hand, this can lead to a waste of resources. On the other hand, it can also lead to not properly protecting the different application components based on their diverse security requirements. To this end, this paper proposes a security solution for cross-cloud applications which is configurable according to the flexible configuration specification given by the devops. Such a specification conforms to a certain UML-based meta-model and is independent of the underlying security tools exploited. In this way, devops can enable to produce a varied security level per each application component that better suits its security requirements. We demonstrate the suitability of our solution through an evaluation showcasing that it can lead to reduced resource consumption without compromising the security of the components that it protects.

Educational testing and learning have evolved from using standard True/False, fill-in-the-blank and multiple choice on paper to more visually enriched formats using interactive multimedia content on digital displays. However, traditional educational application interfaces are primarily mouse-driven, which prevents multiple users working simultaneously. Although touch-based displays have emerged and inspired new developments, they are mainly used in simple tasks. In this paper we show how the multi-touch technology can be extended to collaborative learning and testing at a larger scale, using an existing education implementation for illustration. We propose a Human-Intention-Machine-Interpretation (HIMI) model, which applies a graph-based approach to recognize hand gestures and interpret user intentions. Our focus is not to build a new multi-touch system but to make use of the existing multi-touch technology to enhance learning performance. The HIMI model not only facilitates natural interactions using hand movements on simple tasks, but also supports complex collaborative operations. Our contribution lies in embedding the multi-touch technology in multimedia education, providing a multi-user learning and testing environment which would not have been possible using traditional input devices. We formalize a conceptual model to uniquely interpret user intentions via touch states, state transitions and transition associations. We also propose a set of hand gestures for working with multimedia educational items. User evaluations are conducted to show the feasibility of the proposed hand gestures.

The minimal embedding of the Standard Model in type I string theory is described. The SU(3) color and SU(2) weak interactions arise from two different collections of branes. The correct prediction of the weak angle is obtained for a string scale of 6-8 TeV. Two Higgs doublets are necessary and proton stability is guaranteed. It predicts two massive vector bosons with masses at the TeV scale, as well as a new superweak interaction.

We have used Rossi X-ray Timing Explorer data to measure the lags between soft (2-5 keV) and hard (5-13 keV) photons and to study the aperiodic variability of the superluminal black hole candidate GRS 1915+105 during low-flux states. The power density spectra exhibit quasi-periodic oscillations (QPO) whose frequency increases with increasing count rate and varies in the frequency range 0.6-8 Hz. A correlation between the QPO frequency and the phase lag spectra is reported for the first time. This correlation is found for both the phase lag continuum and the phase lag at the QPO frequency. We find that as the QPO frequency moves to higher values the phase lags reverse sign from positive to negative. The absolute value of the lag always increases with photon energy. The negative (soft) lags are associated with a softer energy spectrum, whereas the positive (hard) lags are seen when the source is harder. We describe a possible scenario that may account for the change in the sign of the lags.

Chapter 25 presents the basic characteristics of Mobile IP, a protocol that allows transparent routing of IP datagrams to mobile nodes on the Internet. The home agent sends datagrams destined for the mobile node through a tunnel to the care-of address. Routing, security and management issues are also discussed.

We revisit the relation between magnetic-field strength ($B$) and gas density ($ρ$) for contracting interstellar clouds and fragments (or, cores), which is central in observationally determining the dynamical importance of magnetic fields in cloud evolution and star formation. Recently, it has been claimed that a relation $B \propto ρ^{2/3} $ is statistically preferred over $B \propto ρ^{1/2}$ in molecular clouds, when magnetic field detections and nondetections from Zeeman observations are combined. This finding has unique observational implications on cloud and core geometry: The relation $B \propto ρ^{2/3} $ can only be realized under spherical contraction. However, no indication of spherical geometry can be found for the objects used in the original statistical analysis of the $B-ρ$ relation. We trace the origin of the inconsistency to simplifying assumptions in the statistical model used to arrive at the $B\propto ρ^{2/3}$ conclusion and to an underestimate of observational uncertainties in the determination of cloud and core densities. We show that, when these restrictive assumptions are relaxed, $B \propto ρ^{1/2}$ is the preferred relation for the (self-gravitating) molecular-cloud data, as theoretically predicted four decades ago.

Some recent observational results impose significant constraints on all the models that have been proposed to explain the Galactic black-hole X-ray sources in the hard state. In particular, it has been found that during the hard state of Cyg X-1 the power-law photon number spectral index is correlated with the average time lag between hard and soft X-rays. Furthermore, the peak frequencies of the four Lorentzians that fit the observed power spectra are correlated with both the photon index and the time lag. We performed Monte Carlo simulations of Compton upscattering of soft, accretion-disk photons in the jet and computed the time lag between hard and soft photons and the power-law index of the resulting photon number spectra. We demonstrate that our jet model naturally explains the above correlations, with no additional requirements and no additional parameters.

Photodeposition of metal catalysts (or their oxides) from their metal ion solutions onto semiconducting oxide supports is an efficient and green preparation method to improve the photocatalytic activity of the latter or the stability of the former when used as electrocatalysts. TiO 2 has been the choice of such a support for metal catalysts since it is inexpensive and stable, but its low electronic conductivity has to be alleviated by either the addition of carbon or photodeposition methodologies that allow for very good metal nanoparticle dispersion on the support nanoparticles, ensuring adequate metal inter-particle contacts and conductivity [1-2]. Graphitic carbon nitride (g-CN), as a semiconductor material with a narrower energy band-gap and higher conductivity than TiO 2 that at the same time shows high stability, is an attractive alternative support for Pt or IrO 2 /Ir (used as hydrogen evolution reaction, HER, and oxygen evolution reaction, OER, electrocatalysts respectively), since it offers stability, tunable conductivity and the opportunity to photodeposit a metal onto it under visible light illumination [3]. In this communication, we report preliminary results for the preparation of Pt/g-CN and IrO 2 /Ir/g-CN supported catalysts by photodeposition of the metals/metal oxides from K 2 PtCl 6 and IrCl 3 solutions onto graphitic nitride nanosheets (g-CNNs); the latter have been prepared by thermal exfoliation of bulk g-CN that has been produced by thermal poly-condensation of melamine. This resulted in supported catalysts with a 30-40% w/w metal catalyst content and a fine dispersion of nanoparticles, especially in the case of Pt (see Figure 1(A) and 1(B)). These materials have been tested as HER and OER electrodes and compared well with commercial Pt/C and IrO 2 electrocatalysts. References [1] Banti, A et al 2024 Molecules 29 2392 [2] Papaderakis, A et al 2020 Catalysts 10 248 [3] Mohd, S et al 2023 Frontiers in Energy Research Acknowledgements: “Greece 2.0” funded by the European Union-NextGenerationEU (H.F.R.I. Project Number: 16443 (H.F.R.I. Project Number: 16443) Figure 1

A handmade metal “tongue” bolted to the inside of a hydrant coupling is a convenient way to divert water from a hydrant. The diverter‐tongue, which takes about an hour to construct, dissipates the energy of the water's flow by directing the flow up, out, and away from the hydrant in a powerful spray.

Be/X-ray binaries are the most numerous group of high-mass X-ray binaries. Their long-term optical and infrared variability reflects the evolution of the circumstellar disk around the luminous companion. This variability manifests photometrically as an excess of flux that increases with wavelength and spectroscopically as line emission. The disk is also expected to generate linear polarization. We present a systematic study of the optical long-term polarimetric variability of Be/X-ray binaries on data collected over 10 years. Our aim is to characterize the polarimetric properties of these systems and to probe the structure of their circumstellar disks. We have been monitoring Be/X-ray binaries visible from the Northern hemisphere with the RoboPol polarimeter. Optical polarimetric variability is a common trait in Be/X-ray binaries. The variability can be attributed to the Be star's circumstellar disk. Our polarization analysis confirms previous claims based on spectroscopic data that the circumstellar disks in BeXBs are, on average, smaller and denser than those in Be stars in non-binary systems. Our data also confirms the presence of highly distorted disks before giant X-ray outbursts, although this result is still affected by the lack of simultaneous and well-sampled observations during major X-ray outbursts.

Understanding how the primary visual cortex of mice represents the external sensory input separately from the internal states is a fundamental challenge in systems neuroscience. Our work contributes to the problem of decoupling the stimulus-driven and internally generated components of neural activity in the primary visual cortex. Internally generated (or intrinsic) activity refers to neural dynamics that are not directly driven by sensory stimuli, reflecting the brain's ongoing, endogenous processes. Neuronal activity encodes both external stimuli and internal cortical states. The internally modulated activity, though not directly observable, can be inferred from the shared structure in population responses, and thus, serves as a proxy for the internal cortical state. We developed a two-phase Partial Least Squares Regression (PLSR) framework that decomposes neural activity into two orthogonal low-dimensional subspaces: (1) a "population" subspace capturing global variability shared across neurons, and (2) a "stimulus" subspace containing dimensions that discriminate between stimulus conditions while being linearly uncorrelated with the population subspace. We focus on the granular (L4) and supragranular (L2/3) layers of awake mice exposed to visual stimuli consisting of optical flow directions, using mesoscopic two-photon calcium imaging. In both L4 and L2/3 layers, many components individually yield above-chance decoding accuracy, yet a small low-dimensional subspace preserves nearly the full decoding performance of the high-dimensional population. Stimulus-driven components exhibit strong cross-mouse correlations, indicating a conserved coding scheme present in both L4 and L2/3. These components are stable across the entire recording session, reflecting robustness of the underlying representation over time. Removing the global modulation did not abolish stimulus discriminability in either layer, suggesting that information about stimulus direction is not dependent on this global signal. Both L4 and L2/3 stimulus components exhibit comparable decoding performance as well as similar tuning representations, suggesting common encoding of stimulus direction across layers.

The author provides an overview of the development of ideas leading to the postulation of photonic gaps in composite materials. He looks at the role of defects and the structure of composite materials, the optimum conditions for gap formation and the fabrication of photonic band gap materials. Finally, future prospects for this field are briefly discussed.

We present Hydra, a blocking synchronization protocol which signi cantly outperforms all previous state-of-the-art synchronization algorithms. We experimentally show that the throughput of Hydra is higher than that of CC-Synch, a state-of-the-art synchronization protocol presented in PPoPP '12, by a multiplicative factor of up to 7.9; this throughput is surprisingly close to the ideal. Hydra implements the combining technique. It also employs user-level threads and schedules them appropriately to ensure their synchronization, as well as to improve performance. We show that by appropriately using user-level threads, we can also get a simple variant of PSim, called PSimX, with highly upgraded performance than that of PSim; PSim is a state-of-the-art wait-free universal construction, rst appeared in SPAA '11. The performance of PSimX, albeit lower than that of Hydra, is also very close to the ideal. Experiments show that by employing user-level threads in other state-of-the-art synchronization protocols, we do not get so big performance advantages as for Hydra and PSimX. To evaluate the performance of Hydra and PSimX, we implement and experimentally evaluate implementations of concurrent queues and stacks, based on Hydra and PSimX. Our queue implementation outperforms by far all current state-of-the-art concurrent<br> queue implementations. Based on our protocols, it is easy to implement useful primitives (like double CAS and others) at a surprisingly low cost.