Bar-Ilan Faculty of Engineering

Federated learning is a technique that allows multiple entities to collaboratively train models using their data without compromising data privacy. However, despite its advantages, federated learning can be susceptible to false data injection attacks. In these scenarios, a malicious entity with control over specific agents in the network can manipulate the learning process, leading to a suboptimal model. Consequently, addressing these data injection attacks presents a significant research challenge in federated learning systems. In this paper, we propose a novel technique to detect and mitigate data injection attacks on federated learning systems. Our mitigation method is a local scheme, performed during a single instance of training by the coordinating node, allowing the mitigation during the convergence of the algorithm. Whenever an agent is suspected to be an attacker, its data will be ignored for a certain period, this decision will often be re-evaluated. We prove that with probability 1, after a finite time, all attackers will be ignored while the probability of ignoring a trustful agent becomes 0, provided that there is a majority of truthful agents. Simulations show that when the coordinating node detects and isolates all the attackers, the model recovers and converges to the truthful model.

We design an adaptive learning algorithm for binary classification problems whose objective is to reduce the cost of misclassified instances derived from the consequences of errors. Our algorithm (Adaptive Cost-Sensitive Learning — AdaCSL) adaptively adjusts the loss function to bridge the difference between the class distributions between subgroups of samples in the training and validation data sets. This adjustment is made for samples with similar predicted probabilities, in such a way that the local cost decreases. This process usually leads to a reduction in cost when applied to the test data set (i.e., local training–test class distributions mismatch). We present empirical evidence that neural networks used with the proposed algorithm yields better cost results on several data sets compared to other approaches. In addition, the proposed AdaCSL algorithm can optimize evaluation metrics other than cost. We …

Weak values and Kirkwood--Dirac (KD) quasiprobability distributions have been independently associated with both foundational issues in quantum theory and advantages in quantum metrology. We propose simple quantum circuits to measure weak values, KD distributions, and spectra of density matrices without the need for post-selection. This is achieved by measuring unitary-invariant, relational properties of quantum states, which are functions of Bargmann invariants, the concept that underpins our unified perspective. Our circuits also enable experimental implementation of various functions of KD distributions, such as out-of-time-ordered correlators (OTOCs) and the quantum Fisher information in post-selected parameter estimation, among others. An upshot is a unified view of nonclassicality in all those tasks. In particular, we discuss how negativity and imaginarity of Bargmann invariants relate to set coherence.

Intraocular pressure (IOP) measurements comprise an essential tool in modern medicine for the early diagnosis of glaucoma, the second leading cause of human blindness. The world's highest prevalence of glaucoma is in low-income countries.Current diagnostic methods require experience in running expensive equipment as well as the use of anesthetic eye drops. We present herein a remote photonic IOP biomonitoring method based on deep learning of secondary speckle patterns, captured by a fast camera, that are reflected from eye sclera stimulated by an external sound wave. By combining speckle pattern analysis with deep learning, high precision measurements are possible.

The significant heterogeneity of Wilms’ tumors between different patients is thought to arise from genetic and epigenetic distortions that occur during various stages of fetal kidney development in a way that is poorly understood. To address this, we characterized the heterogeneity of alternative mRNA splicing in Wilms’ tumors using a publicly available RNAseq dataset of high-risk Wilms’ tumors and normal kidney samples. Through Pareto task inference and cell deconvolution, we found that the tumors and normal kidney samples are organized according to progressive stages of kidney development within a triangle-shaped region in latent space, whose vertices, or “archetypes,” resemble the cap mesenchyme, the nephrogenic stroma, and epithelial tubular structures of the fetal kidney. We identified a set of genes that are alternatively spliced between tumors located in different regions of latent space and found that many of these genes are associated with the Epithelial to Mesenchymal Transition (EMT) and muscle development. Using motif enrichment analysis, we identified putative splicing regulators, some of which are associated with kidney development. Our findings provide new insights into the etiology of Wilms’ tumors and suggest that specific splicing mechanisms in early stages of development may contribute to tumor development in different patients.

The significant heterogeneity of Wilms’ tumors between different patients is thought to arise from genetic and epigenetic distortions that occur during various stages of fetal kidney development in a way that is poorly understood. To address this, we characterized the heterogeneity of alternative mRNA splicing in Wilms’ tumors using a publicly available RNAseq dataset of high-risk Wilms’ tumors and normal kidney samples. Through Pareto task inference and cell deconvolution, we found that the tumors and normal kidney samples are organized according to progressive stages of kidney development within a triangle-shaped region in latent space, whose vertices, or “archetypes,” resemble the cap mesenchyme, the nephrogenic stroma, and epithelial tubular structures of the fetal kidney. We identified a set of genes that are alternatively spliced between tumors located in different regions of latent space and found that many of these genes are associated with the Epithelial to Mesenchymal Transition (EMT) and muscle development. Using motif enrichment analysis, we identified putative splicing regulators, some of which are associated with kidney development. Our findings provide new insights into the etiology of Wilms’ tumors and suggest that specific splicing mechanisms in early stages of development may contribute to tumor development in different patients.

A critical current challenge in the development of all-solid-state lithium batteries (ASSLBs) is reducing the cost of fabrication without compromising the performance. Here we report a sulfide ASSLB based on a high-energy, Co-free LiNiO2 cathode with a robust outside-in structure. This promising cathode is enabled by the high-pressure O2 synthesis and subsequent atomic layer deposition of a unique ultrathin LixAlyZnzOδ protective layer comprising a LixAlyZnzOδ surface coating region and an Al and Zn near-surface doping region. This high-quality artificial interphase enhances the structural stability and interfacial dynamics of the cathode as it mitigates the contact loss and continuous side reactions at the cathode/solid electrolyte interface. As a result, our ASSLBs exhibit a high areal capacity (4.65 mAh cm−2), a high specific cathode capacity (203 mAh g−1), superior cycling stability (92% capacity retention …

Modulation of electronic properties of materials by electric fields is central to the operation of modern semiconductor devices, providing access to complex electronic behaviors and greater freedom in tuning the energy bands of materials. Here, we explore one-dimensional superlattices induced by a confining electrostatic potential in monolayer MoS, a prototypical two-dimensional semiconductor. Using first-principles calculations, we show that periodic potentials applied to monolayer MoS induce electrostatic superlattices in which the response is dominated by structural distortions relative to purely electronic effects. These structural distortions reduce the intrinsic band gap of the monolayer substantially while also polarizing the monolayer through piezoelectric coupling, resulting in spatial separation of charge carriers as well as Stark shifts that produce dispersive minibands. Importantly, these minibands inherit the valley-selective magnetic properties of monolayer MoS, enabling fine control over spin-valley coupling in MoS and similar transition-metal dichalcogenides.

In set-based face recognition, we aim to compute the most discriminative descriptor from an unbounded set of images and videos showing a single person. A discriminative descriptor balances two policies when aggregating information from a given set. The first is a quality-based policy: emphasizing high-quality and down-weighting low-quality images. The second is a diversity-based policy: emphasizing unique images in the set and down-weighting multiple occurrences of similar images as found in video clips which can overwhelm the set representation. This work frames face-set representation as a differentiable coreset selection problem. Our model learns how to select a small coreset of the input set that balances quality and diversity policies using a learned metric parameterized by the face quality, optimized end-to-end. The selection process is a differentiable farthest-point sampling (FPS) realized by approximating the non-differentiable Argmax operation with differentiable sampling from the Gumbel-Softmax distribution of distances. The small coreset is later used as queries in a self and cross-attention architecture to enrich the descriptor with information from the whole set. Our model is order-invariant and linear in the input set size. We set a new SOTA to set face verification on the IJB-B and IJB-C datasets. Our code is publicly available.

Quantum reference frames have attracted renewed interest recently, as their exploration is relevant and instructive in many areas of quantum theory. Among the different types, position and time reference frames have captivated special attention. Here, we introduce and analyze a non-relativistic framework in which each system contains an internal clock, in addition to its external (spatial) degree of freedom and, hence, can be used as a spatiotemporal quantum reference frame. Among other applications of this framework, we show that even in simple scenarios with no interactions, the relative uncertainty between clocks affects the relative spatial spread of the systems.

A universal methodology for coding-decoding the complex amplitude field of an imaged sample in coherent microscopy is presented, where no restrictions on any of the two interferometric beams are required. Thus, the imaging beam can be overlapped with, in general, any other complex amplitude distribution and, in particular, with a coherent and shifted version of itself considering two orthogonal directions. The complex field values are retrieved by a novel Cepstrum-based algorithm, named as Spatial-Shifting Cepstrum (SSC), based on a weighted subtraction of the Cepstrum transform in the cross-correlation term of the object field spectrum in addition with the generation of a complex pupil from the combination of the information retrieved from different holographic recordings (one in horizontal and one in vertical direction) where one of the interferometric beams is shifted 1 pixel. As a result, the field of view is …

A universal methodology for coding-decoding the complex amplitude field of an imaged sample in coherent microscopy is presented, where no restrictions on any of the two interferometric beams are required. Thus, the imaging beam can be overlapped with, in general, any other complex amplitude distribution and, in particular, with a coherent and shifted version of itself considering two orthogonal directions. The complex field values are retrieved by a novel Cepstrum-based algorithm, named as Spatial-Shifting Cepstrum (SSC), based on a weighted subtraction of the Cepstrum transform in the cross-correlation term of the object field spectrum in addition with the generation of a complex pupil from the combination of the information retrieved from different holographic recordings (one in horizontal and one in vertical direction) where one of the interferometric beams is shifted 1 pixel. As a result, the field of view is …

Ordinal classification tasks that require the allocation of limited resources are prevalent in various real-world scenarios. Examples include assessing disease severity in the context of medical resource allocation and categorizing the quality of machines as good, medium, or bad to schedule maintenance treatment within capacity constraints. We propose a comprehensive analytic framework for scenarios that, in addition to including ordinal classification problems, also have constraints on the number of classified samples of classes due to resource limitations. The framework uses a probability matrix generated by a trained ordinal classifier as the input for an optimization model with a minimum misclassification cost objective and resource allocation constraints. We illustrated the equivalence between the formulation of the resource allocation problem into samples and the transportation problem, enabling the utilization …

X-ray imaging is a prevalent technique for non-invasively visualizing the interior of the human body and opaque instruments. In most commercial x-ray modalities, an image is formed by measuring the x-rays that pass through the object of interest. However, despite the potential of scattered radiation to provide additional information about the object, it is often disregarded due to its inherent tendency to cause blurring. Consequently, conventional imaging modalities do not measure or utilize these valuable data. In contrast, we propose and experimentally demonstrate a high-resolution technique for x-ray computed tomography (CT) that measures scattered radiation by exploiting computational ghost imaging (CGI). We show that our method can provide sub-200 {\mu}m resolution, exceeding the capabilities of most existing x-ray imaging modalities. Our research reveals a promising technique for incorporating scattered radiation data in CT scans to improve image resolution and minimize radiation exposure for patients. The findings of our study suggest that our technique could represent a significant advancement in the fields of medical and industrial imaging, with the potential to enhance the accuracy and safety of diagnostic imaging procedures.

We consider a multi-agent system where agents aim to achieve a consensus despite interactions with malicious agents that communicate misleading information. Physical channels supporting communication in cyberphysical systems offer attractive opportunities to detect malicious agents, nevertheless, trustworthiness indications coming from the channel are subject to uncertainty and need to be treated with this in mind. We propose a resilient consensus protocol that incorporates trust observations from the channel and weighs them with a parameter that accounts for how confident an agent is regarding its understanding of the legitimacy of other agents in the network, with no need for the initial observation window that has been utilized in previous works. Analytical and numerical results show that (i) our protocol achieves a resilient consensus in the presence of malicious agents and (ii) the steady-state deviation from nominal consensus can be minimized by a suitable choice of the confidence parameter that depends on the statistics of trust observations.

Photon‐number‐splitting (PNS) is a well‐known theoretical attack on quantum key distribution (QKD) protocols that employ weak coherent states produced by attenuated laser pulses. However, beyond the fact that it has not yet been demonstrated experimentally, its plausibility and effect on quantum bit error rate are questioned. In this work, an experimental scheme is presented for PNS attack employing demonstrated technological capabilities, specifically a single‐photon Raman interaction (SPRINT) in a cavity‐enhanced three‐level atomic system. Several aspects of the proposed implementation are addressed, analytically and simulatively, and the eavesdropper's information gain by the attack is calculated. Furthermore, it is analytically shown that the scheme results in a small (yet non‐zero) quantum bit error rate, and a comparison to purely theoretical analyses in the literature is presented. It is believed that the …

Clipped stochastic gradient descent (SGD) algorithms are among the most popular algorithms for privacy preserving optimization that reduces the leakage of users' identity in model training. This paper studies the convergence properties of these algorithms in a performative prediction setting, where the data distribution may shift due to the deployed prediction model. For example, the latter is caused by strategical users during the training of loan policy for banks. Our contributions are two-fold. First, we show that the straightforward implementation of a projected clipped SGD (PCSGD) algorithm may converge to a biased solution compared to the performative stable solution. We quantify the lower and upper bound for the magnitude of the bias and demonstrate a bias amplification phenomenon where the bias grows with the sensitivity of the data distribution. Second, we suggest two remedies to the bias amplification effect. The first one utilizes an optimal step size design for PCSGD that takes the privacy guarantee into account. The second one uses the recently proposed DiceSGD algorithm [Zhang et al., 2024]. We show that the latter can successfully remove the bias and converge to the performative stable solution. Numerical experiments verify our analysis.

Recent advancements in thin-film lithium niobate (TFLN) photonics have led to a new generation of high-performance electro-optic devices, including modulators, frequency combs, and microwave-to-optical transducers. However, the broader adoption of TFLN-based devices that rely on all-optical nonlinearities have been limited by the sensitivity of quasi-phase matching (QPM), realized via ferroelectric poling, to fabrication tolerances. Here, we propose a scalable fabrication process aimed at improving the wavelength-accuracy of optical frequency mixers in TFLN. In contrast to the conventional pole-before-etch approach, we first define the waveguide in TFLN and then perform ferroelectric poling. This sequence allows for precise metrology before and after waveguide definition to fully capture the geometry imperfections. Systematic errors can also be calibrated by measuring a subset of devices to fine-tune the QPM design for remaining devices on the wafer. Using this method, we fabricated a large number of second harmonic generation devices aimed at generating 737 nm light, with 73% operating within 5 nm of the target wavelength. Furthermore, we also demonstrate thermo-optic tuning and trimming of the devices via cladding deposition, with the former bringing ~96% of tested devices to the target wavelength. Our technique enables the rapid growth of integrated quantum frequency converters, photon pair sources, and optical parametric amplifiers, thus facilitating the integration of TFLN-based nonlinear frequency mixers into more complex and functional photonic systems.

Detecting low concentrations of biomarkers is essential in clinical laboratories. To improve analytical sensitivity, especially in identifying fluorescently labeled molecules, typical optical detection systems, consisting of a photodetector or camera, utilize time-resolved measurements. Taking a different approach, magnetic modulation biosensing (MMB) is a novel technology that combines fluorescently labeled probes and magnetic particles to create a sandwich assay with the target molecules. By concentrating the target molecules and then using time-resolved measurements, MMB provides the rapid and highly sensitive detection of various biomarkers. Here, we propose a novel signal-processing algorithm that enhances the detection and estimation of target molecules at low concentrations. By incorporating both temporally and spatially resolved measurements using human interleukin-8 as a target molecule, we show that the new algorithm provides a 2–4-fold improvement in the limit of detection and an ~25% gain in quantitative resolution.

Often, data stored in memory must be protected from naturally occurring and malicious errors. Methods for constructing codes that are robust with respect to errors injected into the data as well as into the address (in which the data are to be stored) are described. Several ways of extending data-protecting codes to address-and-data protecting codes are presented, and a generalization of the concepts behind CPCs – low cost codes for which no error injected into the data is ever completely masked – is given. A fundamental difference between attacks on the address and data and attacks that only target the data is detailed, and the consequences of this fundamental difference are briefly considered.