Fault Injection Attacks in Spiking Neural Networks and Countermeasures

Spiking Neural Networks (SNN) are fast emerging as an alternative option to Deep Neural Networks (DNN). They are computationally more powerful and provide higher energy-efficiency than DNNs. While exciting at first glance, SNNs contain security-sensitive assets (e.g., neuron threshold voltage) and vulnerabilities (e.g., sensitivity of classification accuracy to neuron threshold voltage change) that can be exploited by the adversaries. We explore global fault injection attacks using external power supply and laser-induced local power glitches on SNN designed using common analog neurons to corrupt critical training parameters such as spike amplitude and neuron’s membrane threshold potential. We also analyze the impact of power-based attacks on the SNN for digit classification task and observe a worst-case classification accuracy degradation of −85.65%. We explore the impact of various design parameters of SNN (e.g., learning rate, spike trace decay constant, and number of neurons) and identify design choices for robust implementation of SNN. We recover classification accuracy degradation by 30–47% for a subset of power-based attacks by modifying SNN training parameters such as learning rate, trace decay constant, and neurons per layer. We also propose hardware-level defenses, e.g., a robust current driver design that is immune to power-oriented attacks, improved circuit sizing of neuron components to reduce/recover the adversarial accuracy degradation at the cost of negligible area, and 25% power overhead. We also propose a dummy neuron-based detection of voltage fault injection at ∼1% power and area overhead each.

Digging into Axion Physics with (Baby)IAXO

Dark matter searches have been ongoing for three decades; the lack of a positive discovery of the main candidate, the WIMP, after dedicated efforts, has put axions and axion-like particles in the spotlight. The three main techniques employed to search for them complement each other well in covering a wide range in the parameter space defined by the axion decay constant and the axion mass. The International AXion Observatory (IAXO) is an international collaboration planning to build the fourth generation axion helioscope, with an unparalleled expected sensitivity and discovery potential. The distinguishing characteristic of IAXO is that it will feature a magnet that is designed to maximise the relevant parameters in sensitivity and which will be equipped with X-ray focusing devices and detectors that have been developed for axion physics. In this paper, we review aspects that motivate IAXO and its prototype, BabyIAXO, in the axion, and ALPs landscape. As part of this Special Issue, some emphasis is given on Spanish participation in the project, of which CAPA (Centro de Astropartículas y Física de Altas Energías of the Universidad de Zaragoza) is a strong promoter.

A study of aftershocks of 20 October 1991 Uttarkashi earthquake

The Uttarkashi earthquake of 20 October 1991, which caused widespread damage in the Galhwal Himalayan region, was followed by a prominent aftershock. activity extending over a period of about two months. The aftershock activity was monitored using temporary networks established after the mainshock and the permanent stations in operation in the region. About 142 aftershocks could be located accurately using the data of these stations. The b-value of the Gutenberg-Richter's relationship for the aftershock sequence works out to be 0.6. The temporal distribution of the aftershocks suggests a hyperbolic decay with a decay constant (p) of 1.17. Macroseismic observations derived from field surveys show good agreement with the instrumentally determined source parameters.

Theoretical aspects on the use of single-time-point dosimetry for radionuclide therapy

Objective: This study considers the error distributions for time-integrated activity (TIA) of single-time-point (STP) methods for patient-specific dosimetry in radionuclide therapy. Approach: The general case with the same pharmaceutical labelled with different radionuclides for imaging and therapy are considered for a mono-exponential time-activity curve. Two methods for STP dosimetry, both based on the combination of one activity estimate with the population-mean effective decay constant, are investigated. The cumulative distribution functions (CDFs) and the probability density functions for the two methods are analytically derived for arbitrary distributions of the biological decay constant. The CDFs are used for determining 95 % coverage intervals of the relative errors for different combinations of imaging time points, physical decay constants, and relative standard deviations of the biological decay constant. Two examples, in the form of kidney dosimetry in [177Lu]Lu-DOTA-TATE therapy and tumour dosimetry for Na[131I]I therapy for thyroid cancer with dosimetry based on imaging of Na[124I]I, are also studied in more detail with analysis of the sensitivity with respect to errors in the mean biological decay constant and to higher moments of the distribution. Main results: The distributions of the relative errors are negatively skewed, potentially leading to the situation that some TIA estimates are highly underestimated even if the majority of estimates are close to the true value. Significance: The main limitation of the studied STP dosimetry methods is thereby the risk of large underestimations of the TIA.

Compositionally Disordered Doped with Cerium Crystalline Garnet Type Materials for Brighter and Faster Scintillations

Ce-doped tetracationic garnets (Gd, M)3Al2Ga3O12(M = Y, Lu) form a family of new multipurpose promising scintillation materials. The aim of this work was to evaluate the scintillation yield in the materials of quaternary garnets activated by cerium ions with partial isovalent substitution of the matrix-forming gadolinium ions by yttrium or lutetium ions.Materials were obtained in the form of polycrystalline ceramic samples, and the best results were shown by samples obtained from the raw materials produced by the coprecipitation method. It was found that ceramics obtained from coprecipitated raw materials ensure a uniform distribution of activator ions in the multi-cationic matrices, which enables the high light yield and fast scintillation kinetics of the scintillation. It was demonstrated that the superstoichiometric content of lutetium/gadolinium in the material is an effective method to suppress phosphorescence accompanied scintillation. For ceramics with the composition (Gd, Lu)3Al2Ga3O12 , a scintillation yield of more than 50.000 ph/MeV was achieved. The scintillation kinetics was measured to be close to the kinetics with a decay constant of 50 ns.In terms of the set of the parameters, the developed scintillation materials are close to the recently developed alkali halide materials LaBr3:Ce, GdBr3:Ce. Moreover, they have high mechanical hardness, are characterized by the absence of hygroscopicity, and are better adapted to the manufacture of pixel detectors used in modern devices for medical diagnostics.

Characteristics of collisional damping of surface ion-acoustic mode in Divertor Plasma Simulator-2 (DiPS-2)

The dissipation of ion-acoustic surface waves propagating in a semi-bounded and collisional plasma which has a boundary with vacuum is theoretically investigated and this result is used for the analysis of edge-relevant plasma simulated by Divertor Plasma Simulator-2 (DiPS-2). The collisional damping of the surface wave is investigated for weakly ionized plasmas by comparing the collisionless Landau damping with the collisional damping as follows: (1) the ratio of ion temperature $({T_i})$ to electron temperature $({T_e})$ should be very small for the weak collisionality $({T_i}/{T_e} \ll 1)$ ; (2) the effect of collisionless Landau damping is dominant for the small parallel wavenumber, and the decay constant is given as $\gamma \approx{-} \sqrt {\mathrm{\pi }/2} {k_\parallel }{\lambda _{De}}\omega _{pi}^2/{\omega _{pe}}$ ; and (3) the collisional damping dominates for the large parallel wavenumber, and the decay constant is given as $\gamma \approx{-} {\nu _{in}}/16$ , where ${\nu _{in}}$ is the ion–neutral collisional frequency. An experimental simulation of the above theoretical prediction has been done in the argon plasma of DiPS-2, which has the following parameters: plasma density ${n_e} = (\textrm{2--9)} \times \textrm{1}{\textrm{0}^{11}}\;\textrm{c}{\textrm{m}^{ - 3}}$ , ${T_e} = 3.7- 3.8\;\textrm{eV}$ , ${T_i} = 0.2- 0.3\;\textrm{eV}$ and collision frequency ${\nu _{in}} = 23- 127\;\textrm{kHz}$ . Although the wavelength should be specified with the given parameters of DiPS-2, the collisional damping is found to be $\gamma = ( - 0.9\;\textrm{to}\; - 5) \times {10^4}\;\textrm{rad}\;{\textrm{s}^{ - 1}}$ for ${k_\parallel }{\lambda _{De}} = 10$ , while the Landau damping is found to be $\gamma = ( - 4\;\textrm{to}\; - 9) \times {10^4}\;\textrm{rad}\;{\textrm{s}^{ - 1}}$ for ${k_\parallel }{\lambda _{De}} = 0.1$ .

Could the GW190814 Secondary Component Be a Bosonic Dark Matter Admixed Compact Star?

We investigate whether the recently observed 2.6 M ⊙ compact object in the gravitational wave event GW190814 can be a bosonic dark matter (DM) admixed compact star. By considering the three constraints of mass, radius, and the stability of such an object, we find that if the DM is made of QCD axions, their particle mass m is constrained to a range that has already been ruled out by the independent constraint imposed by the stellar-mass black hole superradiance process. The 2.6 M ⊙ object can still be a neutron star admixed with at least 2.0 M ⊙ of DM made of axion-like particles (or even a pure axion-like particle star) if 2 × 10−11 eV ≤ m ≤ 2.4 × 10−11 eV (2.9 × 10−11 eV ≤ m ≤ 3.2 × 10−11 eV) with a decay constant of f ≥ 8 × 1017 GeV.

On pion mass and decay constant from theory

We calculate the pion mass from Goldstone modes in the Higgs mechanism related to the neutron decay. The Goldstone pion modes acquire mass by a vacuum misalignment of the Higgs field. The size of the misalignment is controlled by the ratio between the electronic and the nucleonic energy scales. The nucleonic energy scale is involved in the neutron to proton transformation and the electronic scale is involved in the related creation of the electronic state in the course of the electroweak neutron decay. The respective scales influence the mapping of the intrinsic configuration spaces used in our description. The configuration spaces are the Lie groups U(3) for the nucleonic sector and U(2) for the electronic sector. These spaces are both compact and lead to periodic potentials in the Hamiltonians in coordinate space. The periodicity and strengths of these potentials control the vacuum misalignment and leads to a pion mass of 135.2(1.5) MeV with an uncertainty mainly from the fine structure coupling at pionic energies. The pion decay constant 92 MeV results from comparing the fourth order self-coupling in an effective pion field theory with the corresponding fourth order term in the Higgs potential. We suggest analogies with the Goldberger-Treiman relation.