Search

In special relativity, trajectories of particles, whether massive or massless, in 4D, can be displayed in the 3+1 Minkowski space-time manifold. On the other hand, in quantum mechanics, trajectories in phase space are not strictly defined because coordinate and linear momentum cannot be measured simultaneously with arbitrary precision, as these variables do not commute with each other. They are not sharply defined within Hilbert space formalism. Nonetheless, out of the density matrix representing a quantum system the extracted information still yields an enhanced description of its properties, and by arranging adequately the matrix one can acquire additional information from its content. Following these lines of conduct, this paper focuses on a closely related issue, the definition and meaning of velocity and speed of a typical quantum phenomenon, the disentanglement for a bipartite system when its evolution is displayed in a 4D pseudo-space-time, whose coordinates are combinations of the density matrix entries. Formalism is based on the definition of a compact Minkowski manifold, where trajectories are defined using the same reasoning of special relativity in the Minkowski manifold. The space-like and time-like regions acquire different meanings, termed entangled-like and separable-like, respectively. The definition and meaning of velocity and speed of disentanglement follow naturally from the formalism. Depending on the dynamics of the physical system state, trajectories may go forth and back from entanglement to separability regions of the compact Minkowski manifold. When the physical time t is introduced as an intrinsic variable into the formalism, a phenomenon commonly known as sudden death occurs during irreversible evolution when a state that is initially entangled becomes separable., Comment: 16 pages, 7 figures

We consider an asynchronous network of $n$ parties, up to $t$ of which are byzantine, and against $t < \frac{n}{3}$ failures reduce byzantine agreement (BA) on $\ell$-bit inputs to one instance of weak agreement (WA) on $\ell$-bit inputs, one instance of binary BA and $\Theta(\ell n + n^2)$ bits of additional communication. Then, to use in this reduction, we design two information-theoretic WA protocols. In the first one, we use almost-universal hashing to achieve statistical security except with probability $2^{-\lambda}$ against $t < \frac{n}{3}$ failures with $\Theta(n^2(\lambda + \log n + \log \ell)) = \mathcal{O}(\ell n + n^2(\lambda + \log n))$ bits of communication. Then, we modify our first protocol by replacing the hashes with error correcting code symbols and introducing a preliminary step based on the COOL protocol for synchronous multivalued BA [DISC '21] to achieve perfect security against $t \leq \frac{n}{3 + \varepsilon}$ failures with $\Theta(\ell n + n^2)$ bits of communication. Our WA protocols both lead to state-of-the-art information-theoretic extensions from binary BA to multivalued BA., Comment: 20 pages, 0 figures, preprint

The classification of radio-frequency (RF) signals is crucial for applications in robotics, traffic control, and medical devices. Spintronic devices, which respond to RF signals via ferromagnetic resonance, offer a promising solution. Recent studies have shown that a neural network of nanoscale magnetic tunnel junctions can classify RF signals without digitization. However, the complexity of these junctions poses challenges for rapid scaling. In this work, we demonstrate that simple spintronic devices, known as metallic spin-diodes, can effectively perform RF classification. These devices consist of NiFe/Pt bilayers and can implement weighted sums of RF inputs. We experimentally show that chains of four spin-diodes can execute 2x2 pixel filters, achieving high-quality convolutions on the Fashion-MNIST dataset. Integrating the hardware spin-diodes in a software network, we achieve a top-1 accuracy of 88 \% on the first 100 images, compared to 88.4 \% for full software with noise, and 90 \% without noise.

Enumerating the minimal hitting sets of a hypergraph is a problem which arises in many data management applications that include constraint mining, discovering unique column combinations, and enumerating database repairs. Previously, Eiter et al. showed that the minimal hitting sets of an $n$-vertex hypergraph, with treewidth $w$, can be enumerated with delay $O^*(n^{w})$ (ignoring polynomial factors), with space requirements that scale with the output size. We improve this to fixed-parameter-linear delay, following an FPT preprocessing phase. The memory consumption of our algorithm is exponential with respect to the treewidth of the hypergraph., Comment: To appear in ICDT 2025

We consider an asynchronous network of $n$ message-sending parties, up to $t$ of which are byzantine. We study approximate agreement, where the parties obtain approximately equal outputs in the convex hull of their inputs. In their seminal work, Abraham, Amit and Dolev [OPODIS '04] achieve this with the optimal resilience $t < \frac{n}{3}$ with a protocol where each party reliably broadcasts its input every iteration. This takes $\Theta(n^2)$ messages per reliable broadcast, or $\Theta(n^3)$ messages per iteration. In this work, we present optimally resilient asynchronous approximate agreement protocols where we forgo reliable broadcast to require communication proportional to $n^2$ instead of $n^3$. We begin with a protocol for $\omega$-dimensional barycentric agreement with $\mathcal{O}(\omega n^2)$ small messages that does not use reliable broadcast. Then, we achieve edge agreement in a tree of diameter $D$ with $\lceil \log_2 D \rceil$ iterations of a multivalued graded consensus variant. This results in a $\mathcal{O}(\log\frac{1}{\varepsilon})$-round protocol for $\varepsilon$-agreement in $[0, 1]$ with $\mathcal{O}(n^2\log\frac{1}{\varepsilon})$ messages and $\mathcal{O}(n^2\log\frac{1}{\varepsilon}\log\log\frac{1}{\varepsilon})$ bits of communication, improving over the state of the art which matches this complexity only when the inputs are all either $0$ or $1$. Finally, we extend our edge agreement protocol for edge agreement in $\mathbb{Z}$ and thus $\varepsilon$-agreement in $\mathbb{R}$ with quadratic communication, in $\mathcal{O}(\log\frac{M}{\varepsilon})$ rounds where $M$ is the maximum honest input magnitude., Comment: 25 pages, 3 figures, preprint

The ability to process time-series at low energy cost is critical for many applications. Recurrent neural network, which can perform such tasks, are computationally expensive when implementing in software on conventional computers. Here we propose to implement a recurrent neural network in hardware using spintronic oscillators as dynamical neurons. Using numerical simulations, we build a multi-layer network and demonstrate that we can use backpropagation through time (BPTT) and standard machine learning tools to train this network. Leveraging the transient dynamics of the spintronic oscillators, we solve the sequential digits classification task with $89.83\pm2.91~\%$ accuracy, as good as the equivalent software network. We devise guidelines on how to choose the time constant of the oscillators as well as hyper-parameters of the network to adapt to different input time scales., Comment: 7 pages, 4 figures

Virtual Try-On (VTON) is a highly active line of research, with increasing demand. It aims to replace a piece of garment in an image with one from another, while preserving person and garment characteristics as well as image fidelity. Current literature takes a supervised approach for the task, impairing generalization and imposing heavy computation. In this paper, we present a novel zero-shot training-free method for inpainting a clothing garment by reference. Our approach employs the prior of a diffusion model with no additional training, fully leveraging its native generalization capabilities. The method employs extended attention to transfer image information from reference to target images, overcoming two significant challenges. We first initially warp the reference garment over the target human using deep features, alleviating "texture sticking". We then leverage the extended attention mechanism with careful masking, eliminating leakage of reference background and unwanted influence. Through a user study, qualitative, and quantitative comparison to state-of-the-art approaches, we demonstrate superior image quality and garment preservation compared unseen clothing pieces or human figures., Comment: Project page available at https://nadavorzech.github.io/max4zero.github.io/

Current multimodal and multitask foundation models like 4M or UnifiedIO show promising results, but in practice their out-of-the-box abilities to accept diverse inputs and perform diverse tasks are limited by the (usually rather small) number of modalities and tasks they are trained on. In this paper, we expand upon the capabilities of them by training a single model on tens of highly diverse modalities and by performing co-training on large-scale multimodal datasets and text corpora. This includes training on several semantic and geometric modalities, feature maps from recent state of the art models like DINOv2 and ImageBind, pseudo labels of specialist models like SAM and 4DHumans, and a range of new modalities that allow for novel ways to interact with the model and steer the generation, for example image metadata or color palettes. A crucial step in this process is performing discrete tokenization on various modalities, whether they are image-like, neural network feature maps, vectors, structured data like instance segmentation or human poses, or data that can be represented as text. Through this, we expand on the out-of-the-box capabilities of multimodal models and specifically show the possibility of training one model to solve at least 3x more tasks/modalities than existing ones and doing so without a loss in performance. This enables more fine-grained and controllable multimodal generation capabilities and allows us to study the distillation of models trained on diverse data and objectives into a unified model. We successfully scale the training to a three billion parameter model using tens of modalities and different datasets. The resulting models and training code are open sourced at 4m.epfl.ch., Comment: Project page at 4m.epfl.ch

Network measurement involves an inherent tradeoff between accuracy and overhead; higher accuracy typically comes at the expense of greater measurement overhead (measurement frequency, number of probe packets, etc.). Capturing the "right" balance between these two desiderata - high accuracy and low overhead - is a key challenge. However, the manner in which accuracy and overhead are traded off is specific to the measurement method, rendering apples-to-apples comparisons difficult. To address this, we put forth a novel analytical framework for quantifying the accuracy-overhead tradeoff for network measurements. Our framework, inspired by the observer effect in modern physics, introduces the notion of a network observer factor, which formally captures the relation between measurement accuracy and overhead. Using our "network observer framework", measurement methods for the same task can be characterized in terms of their network observer factors, allowing for apples-to-apples comparisons. We illustrate the usefulness of our approach by showing how it can be applied to various application domains and validate its conclusions through experimental evaluation., Comment: This paper is an extended version of "The Observer Effect in Computer Networks", which was published in the ACM Applied Networking Research Workshop (ANRW), 2024

Space solar power (SSP), envisioned for decades as a solution for continuous, stable, and dynamically dispatchable clean energy, has seen tremendous interest and a number of experimental demonstrations in the last few years. A practical implementation has been elusive to date, owing to the high launch costs associated with heavy, rigid photovoltaic (PV) and wireless power transfer (WPT) arrays. Lightweight and flexible solutions for WPT have been demonstrated terrestrially but, to date, have not been deployed and tested in space. In this paper, we present an experimental space demonstration of a lightweight, flexible WPT array powered by custom radio frequency integrated circuits (RFICs). The transmit arrays, receive arrays, and the rest of the system were operated and tested for eight months in Low Earth Orbit (LEO). Results from these experiments, including pointing of the array's beam to Earth and its detection by a ground station, are presented and discussed in detail. Observations and results from this mission uncover existing strengths and weaknesses that inform future steps toward realizing SSP., Comment: 19 pages, 23 figures. These authors contributed equally: Alex Ayling, Austin Fikes, Oren S. Mizrahi, Ailec Wu. Updated on 2024/02/16 to reflect the current and former affiliations of the authors

Aftershocks of aftershocks - and their aftershock cascades - substantially contribute to the increased seismicity rate and the associated elevated seismic hazard after the occurrence of a large earthquake. Current state-of-the-art earthquake forecasting models therefore describe earthquake occurrence using self-exciting point processes, where events can recursively trigger more events according to empirical laws. To estimate earthquake probabilities within future time horizons of interest, a large number of possible realizations of a process are simulated, which is typically associated with long computation times that increase with the desired resolution of the forecast in space, time, or magnitude range. We here propose a machine learning approach to estimate the temporal evolution of the rate of higher-order aftershocks. For this, we train a deep neural network to predict the output of the simulation-based approach, given a parametric description of the rate of direct aftershocks. A comparison of the two approaches reveals that they perform very similarly in describing synthetic datasets generated with the simulation-based approach. Our method has two major benefits over the traditional approach. It is faster by several orders of magnitude, and it is not susceptible to being influenced by the presence or absence of individual `extreme' realizations of the process, and thus enables accurate earthquake forecasting in near-real-time.

Recent advances in large language models (LLMs) have led to the development of various evaluation benchmarks. These benchmarks typically rely on a single instruction template for evaluating all LLMs on a specific task. In this paper, we comprehensively analyze the brittleness of results obtained via single-prompt evaluations across 6.5M instances, involving 20 different LLMs and 39 tasks from 3 benchmarks. To improve robustness of the analysis, we propose to evaluate LLMs with a set of diverse prompts instead. We discuss tailored evaluation metrics for specific use cases (e.g., LLM developers vs. developers interested in a specific downstream task), ensuring a more reliable and meaningful assessment of LLM capabilities. We then implement these criteria and conduct evaluations of multiple models, providing insights into the true strengths and limitations of current LLMs., Comment: Accepted at TACL; pre-MIT Press publication version

Current machine learning models for vision are often highly specialized and limited to a single modality and task. In contrast, recent large language models exhibit a wide range of capabilities, hinting at a possibility for similarly versatile models in computer vision. In this paper, we take a step in this direction and propose a multimodal training scheme called 4M. It consists of training a single unified Transformer encoder-decoder using a masked modeling objective across a wide range of input/output modalities - including text, images, geometric, and semantic modalities, as well as neural network feature maps. 4M achieves scalability by unifying the representation space of all modalities through mapping them into discrete tokens and performing multimodal masked modeling on a small randomized subset of tokens. 4M leads to models that exhibit several key capabilities: (1) they can perform a diverse set of vision tasks out of the box, (2) they excel when fine-tuned for unseen downstream tasks or new input modalities, and (3) they can function as a generative model that can be conditioned on arbitrary modalities, enabling a wide variety of expressive multimodal editing capabilities with remarkable flexibility. Through experimental analyses, we demonstrate the potential of 4M for training versatile and scalable foundation models for vision tasks, setting the stage for further exploration in multimodal learning for vision and other domains., Comment: NeurIPS 2023 Spotlight. Project page at https://4m.epfl.ch/

Simulating tactile perception could potentially leverage the learning capabilities of robotic systems in manipulation tasks. However, the reality gap of simulators for high-resolution tactile sensors remains large. Models trained on simulated data often fail in zero-shot inference and require fine-tuning with real data. In addition, work on high-resolution sensors commonly focus on ones with flat surfaces while 3D round sensors are essential for dexterous manipulation. In this paper, we propose a bi-directional Generative Adversarial Network (GAN) termed SightGAN. SightGAN relies on the early CycleGAN while including two additional loss components aimed to accurately reconstruct background and contact patterns including small contact traces. The proposed SightGAN learns real-to-sim and sim-to-real processes over difference images. It is shown to generate real-like synthetic images while maintaining accurate contact positioning. The generated images can be used to train zero-shot models for newly fabricated sensors. Consequently, the resulted sim-to-real generator could be built on top of the tactile simulator to provide a real-world framework. Potentially, the framework can be used to train, for instance, reinforcement learning policies of manipulation tasks. The proposed model is verified in extensive experiments with test data collected from real sensors and also shown to maintain embedded force information within the tactile images.

Spintronic devices have recently attracted a lot of attention in the field of unconventional computing due to their non-volatility for short and long term memory, non-linear fast response and relatively small footprint. Here we report how voltage driven magnetization dynamics of dual free layer perpendicular magnetic tunnel junctions enable to emulate spiking neurons in hardware. The output spiking rate was controlled by varying the dc bias voltage across the device. The field-free operation of this two terminal device and its robustness against an externally applied magnetic field make it a suitable candidate to mimic neuron response in a dense Neural Network (NN). The small energy consumption of the device (4-16 pJ/spike) and its scalability are important benefits for embedded applications. This compact perpendicular magnetic tunnel junction structure could finally bring spiking neural networks (SNN) to sub-100nm size elements.

Teleoperation enables a user to perform dangerous tasks (e.g., work in disaster zones or in chemical plants) from a remote location. Nevertheless, common approaches often provide cumbersome and unnatural usage. In this letter, we propose TeleFMG, an approach for teleoperation of a multi-finger robotic hand through natural motions of the user's hand. By using a low-cost wearable Force-Myography (FMG) device, musculoskeletal activities on the user's forearm are mapped to hand poses which, in turn, are mimicked by a robotic hand. The mapping is performed by a spatio-temporal data-based model based on the Temporal Convolutional Network. The model considers spatial positions of the sensors on the forearm along with temporal dependencies of the FMG signals. A set of experiments show the ability of a teleoperator to control a multi-finger hand through intuitive and natural finger motion. A robot is shown to successfully mimic the user's hand in object grasping and gestures. Furthermore, transfer to a new user is evaluated while showing that fine-tuning with a limited amount of new data significantly improves accuracy.

Large apertures in space are critical for high-power and high-bandwidth applications spanning wireless power transfer (WPT) and communication, however progress on this front is stunted by the geometric limitations of rocket flight. Here, we present a light and flexible 10GHz array, which is composed of dipole antennas co-cured to a glass-fiber composite. The arrays are designed to dynamically conform to new shapes and to be flexible enough to fold completely flat, coil, and pop back up upon deployment. The design was chosen to be amenable to scalable, automated manufacturing - a requirement for the massive production necessary for large apertures. Moreover, the arrays passed the standard gamut of required space-qualification testing: the antennas can survive mechanical stress, extreme temperatures, high-frequency temperature cycling, and prolonged stowage in the flattened configuration. The elements exhibit excellent electromagnetic performance - with a return ratio better than -10dB over a bandwidth of 1.5GHz and a single lobe half-power beam width of greater than $110^\circ$ suitable for broad range beamforming and with excellent manufacturing consistency. Moreover, its mechanical durability vis-a-vis extreme temperatures and protracted stowage lends itself to demanding space applications. This lightweight and scalable array is equipped to serve a host of new space-based radio-frequency technologies and applications which leverage large, stowable and durable array apertures., Comment: 12 pages, 21 figures

The Invertible Bloom Lookup Table (IBLT) is a probabilistic data structure for set representation, with applications in network and traffic monitoring. It is known for its ability to list its elements, an operation that succeeds with high probability for sufficiently large table. However, listing can fail even for relatively small sets. This paper extends recent work on the worst-case analysis of IBLT, which guarantees successful listing for all sets of a certain size, by introducing more general IBLT schemes. These schemes allow for greater freedom in the implementation of the insert, delete, and listing operations and demonstrate that the IBLT memory can be reduced while still maintaining successful listing guarantees. The paper also explores the time-memory trade-off of these schemes, some of which are based on linear codes and \(B_h\)-sequences over finite fields.

Adversarial attacks on deep-learning models have been receiving increased attention in recent years. Work in this area has mostly focused on gradient-based techniques, so-called "white-box" attacks, wherein the attacker has access to the targeted model's internal parameters; such an assumption is usually unrealistic in the real world. Some attacks additionally use the entire pixel space to fool a given model, which is neither practical nor physical (i.e., real-world). On the contrary, we propose herein a direct, black-box, gradient-free method that uses the learned image manifold of a pretrained generative adversarial network (GAN) to generate naturalistic physical adversarial patches for object detectors. To our knowledge this is the first and only method that performs black-box physical attacks directly on object-detection models, which results with a model-agnostic attack. We show that our proposed method works both digitally and physically. We compared our approach against four different black-box attacks with different configurations. Our approach outperformed all other approaches that were tested in our experiments by a large margin., Comment: Accepted at MLCS @ ECML-PKDD 2024

Avoiding the use of mathematical formalism, this essay exposes the quantum mechanics the phenomenon of non-locality in terms of a metaphor involving human twins focused on their hands' dexterity attribute., Comment: 12 pages

Phased arrays have enabled advances in communications, sensing, imaging, and wireless power transfer. In all these applications, large apertures enable higher power, higher data rates, higher resolution, and complex functionalities, but are elusive owing to a correspondingly large cost, mass, and physical size. Flexible phased arrays (FPAs) show potential in breaking this trade-off. Their thinness and extremely low mass allow FPAs to be folded, rolled, or otherwise compressed into smaller sizes, thus enabling new regimes of transport and entirely new applications currently not possible. Though a number laboratory prototypes of FPAs have been constructed, the economics of large-scale FPA production has yet to be explored. This paper presents a model FPA architecture and a cost model for producing it at large-scale. The estimate of the per-unit-area cost is bounded by a three-tiered approach. The cost model projects a "middle" estimate for FPA production at $89 per square meter. Estimates for aerial mass density and startup cost are also discussed. This cost model demonstrates that an FPA can be produced at an efficient price point and can potentially replace existing solutions for space, communications, and vehicular applications that demand lightweight, portability, and durability in extreme conditions.

Spintronic nano-oscillators with reduced non-linearity could offer key benefits for realizing neuromorphic applications such as spike-based neurons and frequency multiplexing in neural networks. Here, we experimentally demonstrate the reduction in non-linearity of a spin-Hall nano-oscillator (SHNO) by compensation of its effective magnetic anisotropy. The study involves optimization of Co/Ni multilayer growth to achieve the compensation, followed by spin diode measurements on patterned microstrips to quantify their anisotropy. The relation between the second ($H_{k2}$ = 0.47 mT) and the first order ($H_{k1}^{eff}$ = $-$0.8 mT) anisotropy fields reveals the existence of an easy cone, thereby validating the presence of compensation. Furthermore, we demonstrate a synapse based on the compensated spin diode which has a fixed frequency when the input power is varied. We then study the current-induced auto-oscillation properties of SHNOs on compensated films by patterning nano-constrictions of widths 200 and 100 nm. The invariance of the resonance frequency and linewidth of the compensated SHNO with applied dc current indicates the absence of non-linearity. This independence is maintained irrespective of the applied external fields and its orientations. The compensated SHNO obtained has a linewidth of 1.1 MHz and a peak output power of up to 1 pW/MHz emulating a nano-neuron with a low linewidth and a fixed frequency.

The Invertible Bloom Lookup Table (IBLT) is a probabilistic concise data structure for set representation that supports a listing operation as the recovery of the elements in the represented set. Its applications can be found in network synchronization and traffic monitoring as well as in error-correction codes. IBLT can list its elements with probability affected by the size of the allocated memory and the size of the represented set, such that it can fail with small probability even for relatively small sets. While previous works only studied the failure probability of IBLT, this work initiates the worst case analysis of IBLT that guarantees successful listing for all sets of a certain size. The worst case study is important since the failure of IBLT imposes high overhead. We describe a novel approach that guarantees successful listing when the set satisfies a tunable upper bound on its size. To allow that, we develop multiple constructions that are based on various coding techniques such as stopping sets and the stopping redundancy of error-correcting codes, Steiner systems, and covering arrays as well as new methodologies we develop. We analyze the sizes of IBLTs with listing guarantees obtained by the various methods as well as their mapping memory consumption. Lastly, we study lower bounds on the achievable sizes of IBLT with listing guarantees and verify the results in the paper by simulations.