arXiv subjects

Noa Zilberman

Publications and source records attributed to Noa Zilberman.

At least 19 recordsLinked to original sources

arXiv2026

Improving the Energy Efficiency of High Throughput Computing: A Measurement-Based Case Study

The significant energy consumed by data centers has become a concern both for costs and associated carbon emissions. In particular, the energy efficiency of servers is a key consideration for data center operators, and understanding servers' power consumption under different operating conditions is an important aspect of it. In this paper, we present a measurement-based case study of high-throughput computing. We analyze power usage information of an operational data center, combined with focused measurements of power reduction techniques for a representative high-throughput workload. The study points out the obstacles encountered by data center operators in their efforts to minimize energy consumption and carbon emissions, and discusses the impact of server configuration adjustments on the energy consumption of processing jobs. We offer actionable recommendations for decreasing the energy usage of servers, while considering both performance and carbon emissions.

cs.NI↗

arXiv2026

In-Network Market Prediction Using Machine Learning and Limit Order Books

Machine learning is significantly transforming algorithmic trading, yet the requirement for rapid execution speeds persists. While both aspects aim to boost profitability, embedding advanced machine-learning techniques with reduced trading latency presents a notable challenge. Adopting in-network machine learning, which involves offloading inference to programmable network devices, offers a delicate equilibrium in this trade-off. In this paper, we present LOBIN, a solution that utilizes machine learning within the network for market prediction based on high-frequency market data feeds. LOBIN is adept at constructing limit order books and performing inference directly within programmable switches. When compared to server-based benchmarks, LOBIN not only predicts future stock price movements with higher throughput but also maintains robust machine learning performance. It achieves over a 10% reduction in latency compared to the NASDAQ order-matching server benchmark and delivers microsecond-level latency. Furthermore, the machine learning performance of LOBIN can be further enhanced through the adoption of a hybrid deployment approach that integrates both the switch and the servers. Our evaluation demonstrates that among all data feeds of evaluated stocks, the application of hybrid deployment results in approximately 45% of the traffic and 38\% of the total potential transaction value being processed within switches without server intervention, reducing latency while ensuring that the average change in error rate of predictions remains at around 3% relative to benchmarks based solely on server use.

cs.NI↗

arXiv2026

Quantum fluxes and $\langle\hat{\Phi}^2\rangle$ for a non-minimally coupled scalar field: ringdown and tail on approaching the polar Kerr inner horizon

We compute $\langle\hat{\Phi}^{2}\rangle_\text{ren}$ as well as the energy fluxes $\langle \hat{T}_{uu}\rangle_\text{ren}$ and $\langle \hat{T}_{vv}\rangle_\text{ren}$ (where $u$ and $v$ are the standard Eddington-Finkelstein coordinates) associated with a quantum massless real scalar field $\hat{\Phi}$, with a general curvature coupling constant $\xi$, near the inner horizon (IH) of a Kerr black hole, along the axis of rotation. The quantum field is in the Unruh state, corresponding to an evaporating black hole. We renormalize these quantities by the state-subtraction method. We drop the assumption of minimal coupling to the curvature, thereby generalizing the results of arXiv:2203.08502 for the fluxes at the IH. This requires understanding the asymptotic behavior of $\langle\hat{\Phi}^{2}\rangle_\text{ren}$ neat the IH. State subtraction allows us to push the computation of $\langle\hat{\Phi}^{2}\rangle_\text{ren}$ along the axis of rotation in the Kerr interior in arXiv:2409.17464 deeper into the near-IH region, exposing their final asymptotic behavior on approaching the IH. For $\langle\hat{\Phi}^{2}\rangle_\text{ren}$ (a $\xi$-independent quantity in the Kerr case), we find that the approach to its finite asymptotic IH value is given, per $\ell$-mode, by a ringdown phase (namely exponentially damped oscillations), followed by an inverse-power tail, both in the tortoise coordinate $r_{*}$ (which diverges at the IH). Interestingly, in the regime where the ringing dominates, the ringing's complex frequencies are (numerically) found to match twice the well-known classical quasinormal-mode frequencies in Kerr, and the inverse-power tails are found to be $r_{*}^{-2\ell-3}$ (resembling Price's law in the classical black hole exterior, upon replacement $t\to r_*$). [Abridged]

gr-qc↗

arXiv2025

Detecting Anomalies in Machine Learning Infrastructure via Hardware Telemetry

Modern machine learning (ML) has grown into a tightly coupled, full-stack ecosystem that combines hardware, software, network, and applications. Many users rely on cloud providers for elastic, isolated, and cost-efficient resources. Unfortunately, these platforms as a service use virtualization, which means operators have little insight into the users' workloads. This hinders resource optimizations by the operator, which is essential to ensure cost efficiency and minimize execution time. In this paper, we argue that workload knowledge is unnecessary for system-level optimization. We propose Reveal, which takes a hardware-centric approach, relying only on hardware signals - fully accessible by operators. Using low-level signals collected from the system, Reveal detects anomalies through an unsupervised learning pipeline. The pipeline is developed by analyzing over 30 popular ML models on various hardware platforms, ensuring adaptability to emerging workloads and unknown deployment patterns. Using Reveal, we successfully identified both network and system configuration issues, accelerating the DeepSeek model by 5.97%.

cs.PF↗

arXiv2025

Computation of the semiclassical outflux emerging from a collapsing spherical null shell

We consider a minimally coupled, massless quantum scalar field $\hat{\Phi}$ propagating in the background geometry of a four-dimensional black hole formed by the collapse of a spherical thin null shell, with a Minkowski interior and a Schwarzschild exterior. The field is taken in the natural ``in'' vacuum state, namely, the quantum state in which no excitations arrive from past null infinity. Within the semiclassical framework, we analyze the vacuum polarization $\left\langle\hat{\Phi}^{2}\right\rangle _{\text{ren}}$ and the energy outflux density $\left\langle \hat{T}_{uu}\right\rangle _{\text{ren}}$ (where $u$ is the standard null Eddington coordinate) just outside the shell. Using the point-splitting method, we derive closed-form analytical expressions for both these semiclassical quantities. In particular, our result for $\left\langle \hat{T}_{uu}\right\rangle _{\text{ren}}$ reveals that it vanishes like $(1-2M_{0}/r_{0})^{2}$ as the shell collapses toward the event horizon, where $M_{0}$ is the shell's mass and $r_{0}$ is the value of the area coordinate $r$ at the evaluation point. This confirms that, along a late-time outgoing null geodesic (i.e., one that emerges from the shell very close to the event horizon and propagates toward future null infinity), the outflux gradually evolves (from virtually zero) up to its final Hawking-radiation value while the geodesic traverses the strong-field region (rather than the Hawking-radiation outflux being emitted entirely from the collapsing shell, which would lead to significant backreaction effects).

gr-qc↗

arXiv2024

Computation of $\langle \Phi^2\rangle$ and quantum fluxes at the polar interior of a spinning black hole

Renormalization of physical quantities for quantum field theories in curved spacetimes can be achieved via the consistent subtraction of counterterms within a regularization scheme such as a point-splitting method. Pragmatic mode-sum regularization (PMR) is a point-splitting method which is particularly suitable for rotating black hole spacetimes. We extend and tailor the t-splitting variant of PMR specifically for the interior of a Kerr black hole on the axis of rotation, focusing on a minimally-coupled massless scalar field in the physically-motivated Unruh state. The method addresses unique challenges in the black hole interior that do not occur outside. In particular, while the infinite sum over multipolar number l converges in the black hole exterior, it diverges in the interior, necessitating the subtraction of a so-called intermediate divergence which includes introducing an additional "small" split in the direction of the polar angle. This procedure is outlined and justified, along with the standard PMR method's counterterms subtraction. We apply this method to calculate the renormalized energy-momentum fluxes $\langle T_{uu}\rangle^U_\text{ren}$, $\langle T_{vv}\rangle^U_\text{ren}$ (where u and v are the standard Eddington coordinates) and the renormalized field square $\langle \Phi^2\rangle^U_\text{ren}$ throughout the black hole interior, spanning from (just off) the event horizon to (just off) the inner horizon. Special emphasis is placed on the inner horizon vicinity, where our t-splitting results for the fluxes asymptote to those obtained directly at the inner horizon using a different method in a previous work. In an Appendix, we develop an alternative t-splitting PMR variant which does not include the intermediate divergence subtraction. We utilize it for independent computations that are used to verify the standard t-splitting variant presented in the main text.

gr-qc↗

arXiv2024

Governing Through the Cloud: The Intermediary Role of Compute Providers in AI Regulation

As jurisdictions around the world take their first steps toward regulating the most powerful AI systems, such as the EU AI Act and the US Executive Order 14110, there is a growing need for effective enforcement mechanisms that can verify compliance and respond to violations. We argue that compute providers should have legal obligations and ethical responsibilities associated with AI development and deployment, both to provide secure infrastructure and to serve as intermediaries for AI regulation. Compute providers can play an essential role in a regulatory ecosystem via four key capacities: as securers, safeguarding AI systems and critical infrastructure; as record keepers, enhancing visibility for policymakers; as verifiers of customer activities, ensuring oversight; and as enforcers, taking actions against rule violations. We analyze the technical feasibility of performing these functions in a targeted and privacy-conscious manner and present a range of technical instruments. In particular, we describe how non-confidential information, to which compute providers largely already have access, can provide two key governance-relevant properties of a computational workload: its type-e.g., large-scale training or inference-and the amount of compute it has consumed. Using AI Executive Order 14110 as a case study, we outline how the US is beginning to implement record keeping requirements for compute providers. We also explore how verification and enforcement roles could be added to establish a comprehensive AI compute oversight scheme. We argue that internationalization will be key to effective implementation, and highlight the critical challenge of balancing confidentiality and privacy with risk mitigation as the role of compute providers in AI regulation expands.

cs.CY↗

arXiv2022

Quantum fluxes at the inner horizon of a spinning black hole

Rotating or charged classical black holes in isolation possess a special surface in their interior, the Cauchy horizon, beyond which the evolution of spacetime (based on the equations of General Relativity) ceases to be deterministic. In this work, we study the effect of a quantum massless scalar field on the Cauchy horizon inside a rotating (Kerr) black hole that is evaporating via the emission of Hawking radiation (corresponding to the field being in the Unruh state). We calculate the flux components (in Eddington coordinates) of the renormalized stress-energy tensor of the field on the Cauchy horizon, as functions of the black hole spin and of the polar angle. We find that these flux components are generically non-vanishing. Furthermore, we find that the flux components change sign as these parameters vary. The signs of the fluxes are important, as they provide an indication of whether the Cauchy horizon expands or crushes (when backreaction is taken into account). Regardless of these signs, our results imply that the flux components generically diverge on the Cauchy horizon when expressed in coordinates which are regular there. This is the first time that irregularity of the Cauchy horizon under a semiclassical effect is conclusively shown for (four-dimensional) spinning black holes.

gr-qc↗

arXiv2022

IIsy: Practical In-Network Classification

The rat race between user-generated data and data-processing systems is currently won by data. The increased use of machine learning leads to further increase in processing requirements, while data volume keeps growing. To win the race, machine learning needs to be applied to the data as it goes through the network. In-network classification of data can reduce the load on servers, reduce response time and increase scalability. In this paper, we introduce IIsy, implementing machine learning classification models in a hybrid fashion using off-the-shelf network devices. IIsy targets three main challenges of in-network classification: (i) mapping classification models to network devices (ii) extracting the required features and (iii) addressing resource and functionality constraints. IIsy supports a range of traditional and ensemble machine learning models, scaling independently of the number of stages in a switch pipeline. Moreover, we demonstrate the use of IIsy for hybrid classification, where a small model is implemented on a switch and a large model at the backend, achieving near optimal classification results, while significantly reducing latency and load on the servers.

cs.NI↗

arXiv2022

Automating In-Network Machine Learning

Using programmable network devices to aid in-network machine learning has been the focus of significant research. However, most of the research was of a limited scope, providing a proof of concept or describing a closed-source algorithm. To date, no general solution has been provided for mapping machine learning algorithms to programmable network devices. In this paper, we present Planter, an open-source, modular framework for mapping trained machine learning models to programmable devices. Planter supports a wide range of machine learning models, multiple targets and can be easily extended. The evaluation of Planter compares different mapping approaches, and demonstrates the feasibility, performance, and resource efficiency for applications such as anomaly detection, financial transactions, and quality of experience. The results show that Planter-based in-network machine learning algorithms can run at line rate, have a negligible effect on latency, coexist with standard switching functionality, and have no or minor accuracy trade-offs.

cs.NI↗

arXiv2022

Quantum fluxes at the inner horizon of a near-extremal spherical charged black hole

We analyze and compute the semiclassical stress-energy flux components, the outflux $\langle T_{uu}\rangle$ and the influx $\langle T_{vv}\rangle$ ($u$ and $v$ being the standard null Eddington coordinates), at the inner horizon (IH) of a Reissner-Nordström black hole (BH) of mass $M$ and charge $Q$, in the near-extremal domain in which $Q/M$ approaches $1$. We consider a minimally-coupled massless quantum scalar field, in both Hartle-Hawking ($H$) and Unruh ($U$) states, the latter corresponding to an evaporating BH. The near-extremal domain lends itself to an analytical treatment which sheds light on the behavior of various quantities on approaching extremality. We explore the behavior of the three near-IH flux quantities $\langle T_{uu}^-\rangle^U$, $\langle T_{vv}^-\rangle^U$, and $\langle T_{uu}^-\rangle^H=\langle T_{vv}^-\rangle^H$, as a function of the small parameter $Δ\equiv\sqrt{1-(Q/M)^2}$ (where the superscript "$-$" refers to the IH value). We find that in the near-extremal domain $\langle T_{uu}^-\rangle^U\cong\langle T_{uu}^-\rangle^H=\langle T_{vv}^-\rangle^H$ behaves as $\proptoΔ^5$. In contrast, $\langle T_{vv}^-\rangle^U$ behaves as $\proptoΔ^4$, and we calculate the prefactor analytically. It therefore follows that the semiclassical fluxes at the IH neighborhood of an evaporating near-extremal spherical charged BH are dominated by the influx $\langle T_{vv}\rangle^U$. In passing, we also find an analytical expression for the transmission coefficient outside a Reissner-Nordström BH to leading order in small frequencies (which turns out to be a crucial ingredient of our near-extremal analysis). Furthermore, we explicitly obtain the near-extremal Hawking-evaporation rate ($\proptoΔ^4$), with an analytical expression for the prefactor (obtained here for the first time to the best of our knowledge). [Abridged]

gr-qc↗

arXiv2022

Two-point function of a quantum scalar field in the interior region of a Kerr black hole

Quantum field effects on a classical background spacetime may be obtained from the semiclassical equations of General Relativity with the expectation value of the stress-energy tensor of the quantum field as a source. This expectation value can be calculated from Hadamard's elementary two-point function, which in practice is given in terms of sums of products of field modes evaluated at two spacetime points. We derive expressions for the two-point function for a massless scalar field in the Unruh state on a Kerr black hole spacetime. Our main result in this paper is a novel expression valid when the two points lie inside the black hole; we also (re-)derive, using a new method, the known expression valid when the two points lie outside the black hole. We achieve these expressions by finding relationships between Unruh modes, defined in terms of the retarded Kruskal coordinate, and Eddington modes, defined in terms of the Eddington coordinates. While our starting expression for the two-point function is written in terms of the Unruh modes, we give our final expression in terms of the Eddington modes, which have the computational advantage that they decompose into factors that obey ordinary differential equations. In an appendix we also derive expressions for the bare mode contributions to the flux components of the stress-energy tensor for a minimally-coupled massless scalar field inside the black hole. Our results thus lay the groundwork for future calculations of quantum effects inside a Kerr black hole.

gr-qc↗

arXiv2021

Filling gaps in trustworthy development of AI

The range of application of artificial intelligence (AI) is vast, as is the potential for harm. Growing awareness of potential risks from AI systems has spurred action to address those risks, while eroding confidence in AI systems and the organizations that develop them. A 2019 study found over 80 organizations that published and adopted "AI ethics principles'', and more have joined since. But the principles often leave a gap between the "what" and the "how" of trustworthy AI development. Such gaps have enabled questionable or ethically dubious behavior, which casts doubts on the trustworthiness of specific organizations, and the field more broadly. There is thus an urgent need for concrete methods that both enable AI developers to prevent harm and allow them to demonstrate their trustworthiness through verifiable behavior. Below, we explore mechanisms (drawn from arXiv:2004.07213) for creating an ecosystem where AI developers can earn trust - if they are trustworthy. Better assessment of developer trustworthiness could inform user choice, employee actions, investment decisions, legal recourse, and emerging governance regimes.

cs.AI↗

arXiv2021

Analysis of quantum effects inside spherical charged black holes

We numerically compute the renormalized expectation value $\langle\hatΦ^{2}\rangle_{ren}$ of a minimally-coupled massless quantum scalar field in the interior of a four-dimensional Reissner-Nordstrom black hole, in both the Hartle-Hawking and Unruh states. To this end we use a recently developed mode-sum renormalization scheme based on covariant point splitting. In both quantum states, $\langle\hatΦ^{2}\rangle_{ren}$ is found to approach a \emph{finite} value at the inner horizon (IH). The final approach to the IH asymptotic value is marked by an inverse-power tail $r_{*}^{-n}$, where $r_{*}$ is the Regge-Wheeler "tortoise coordinate", and with $n=2$ for the Hartle-Hawking state and $n=3$ for the Unruh state. We also report here the results of an analytical computation of these inverse-power tails of $\langle\hatΦ^{2}\rangle_{ren}$ near the IH. Our numerical results show very good agreement with this analytical derivation (for both the power index and the tail amplitude), in both quantum states. Finally, from this asymptotic behavior of $\langle\hatΦ^{2}\rangle_{ren}$ we analytically compute the leading-order asymptotic behavior of the trace $\langle\hat{T}_μ^μ\rangle_{ren}$ of the renormalized stress-energy tensor at the IH. In both quantum states this quantity is found to diverge like $b(r-r_{-})^{-1}r_{*}^{-n-2}$ (with $n$ specified above, and with a known parameter $b$). To the best of our knowledge, this is the first fully-quantitative derivation of the asymptotic behavior of these renormalized quantities at the inner horizon of a four-dimensional Reissner-Nordstrom black hole.

gr-qc↗

arXiv2020

Quantum fluxes at the inner horizon of a spherical charged black hole

In an ongoing effort to explore quantum effects on the interior geometry of black holes, we explicitly compute the semiclassical flux components $\left\langle T_{uu}\right\rangle _{ren}$ and $\left\langle T_{vv}\right\rangle _{ren}$ ($u$ and $v$ being the standard Eddington coordinates) of the renormalized stress-energy tensor for a minimally-coupled massless quantum scalar field, in the vicinity of the inner horizon (IH) of a Reissner-Nordström black hole. These two flux components seem to dominate the effect of backreaction in the IH vicinity; and furthermore, their regularization procedure reveals remarkable simplicity. We consider the Hartle-Hawking and Unruh quantum states, the latter corresponding to an evaporating black hole. In both quantum states, we compute $\left\langle T_{uu}\right\rangle _{ren}$ and $\left\langle T_{vv}\right\rangle _{ren}$ in the IH vicinity for a wide range of $Q/M$ values. We find that both $\left\langle T_{uu}\right\rangle _{ren}$ and $\left\langle T_{vv}\right\rangle _{ren}$ attain finite asymptotic values at the IH. Depending on $Q/M$, these asymptotic values are found to be either positive or negative (or vanishing in-between). Note that having a nonvanishing $\left\langle T_{vv}\right\rangle _{ren}$ at the IH implies the formation of a curvature singularity on its ingoing section, the Cauchy horizon. Motivated by these findings, we also take initial steps in the exploration of the backreaction effect of these semiclassical fluxes on the near-IH geometry.

gr-qc↗

arXiv2020

Toward Trustworthy AI Development: Mechanisms for Supporting Verifiable Claims

With the recent wave of progress in artificial intelligence (AI) has come a growing awareness of the large-scale impacts of AI systems, and recognition that existing regulations and norms in industry and academia are insufficient to ensure responsible AI development. In order for AI developers to earn trust from system users, customers, civil society, governments, and other stakeholders that they are building AI responsibly, they will need to make verifiable claims to which they can be held accountable. Those outside of a given organization also need effective means of scrutinizing such claims. This report suggests various steps that different stakeholders can take to improve the verifiability of claims made about AI systems and their associated development processes, with a focus on providing evidence about the safety, security, fairness, and privacy protection of AI systems. We analyze ten mechanisms for this purpose--spanning institutions, software, and hardware--and make recommendations aimed at implementing, exploring, or improving those mechanisms.

cs.CY↗

arXiv2019

Partitioned Paxos via the Network Data Plane

Consensus protocols are the foundation for building fault-tolerant, distributed systems, and services. They are also widely acknowledged as performance bottlenecks. Several recent systems have proposed accelerating these protocols using the network data plane. But, while network-accelerated consensus shows great promise, current systems suffer from an important limitation: they assume that the network hardware also accelerates the application itself. Consequently, they provide a specialized replicated service, rather than providing a general-purpose high-performance consensus that fits any off-the-shelf application. To address this problem, this paper proposes Partitioned Paxos, a novel approach to network-accelerated consensus. The key insight behind Partitioned Paxos is to separate the two aspects of Paxos, agreement, and execution, and optimize them separately. First, Partitioned Paxos uses the network forwarding plane to accelerate agreement. Then, it uses state partitioning and parallelization to accelerate execution at the replicas. Our experiments show that using this combination of data plane acceleration and parallelization, Partitioned Paxos is able to provide at least x3 latency improvement and x11 throughput improvement for a replicated instance of a RocksDB key-value store.

cs.DC↗

arXiv2018

LaKe: An Energy Efficient, Low Latency, Accelerated Key-Value Store

Key-value store is a popular type of cloud computing applications. The performance of key-value store applications have been shown to be very sensitive to load within the data center, and in particular to latency. As load within data center increases, it is becoming hard to maintain key-value store applications' performance, without exceeding both the processing capacity of hosts and the power budgets of racks. In this paper, we present LaKe: a low latency, power efficient key-value store design for cloud applications. LaKe is a modular design, combining multiple cores and cache layering, both in hardware and software. LaKe achieves full line rate throughput, while maintaining a latency of 1.1us and better power efficiency than existing hardware based memcached designs. Using the modularity of our design, we study trade-offs in the use of on-chip memory, SRAM and DRAM in accelerated designs and provide insights for future architectures.

cs.AR↗