Search

Traditional semi-supervised object detection methods assume a fixed set of object classes (in-distribution or ID classes) during training and deployment, which limits performance in real-world scenarios where unseen classes (out-of-distribution or OOD classes) may appear. In such cases, OOD data is often misclassified as ID, thus harming the ID classes accuracy. Open-set methods address this limitation by filtering OOD data to improve ID performance, thereby limiting the learning process to ID classes. We extend this to a more natural open-world setting, where the OOD classes are not only detected but also incorporated into the learning process. Specifically, we explore two key questions: 1) how to accurately detect OOD samples, and, most importantly, 2) how to effectively learn from the OOD samples in a semi-supervised object detection pipeline without compromising ID accuracy. To address this, we introduce an ensemble-based OOD Explorer for detection and classification, and an adaptable semi-supervised object detection framework that integrates both ID and OOD data. Through extensive evaluation on different open-world scenarios, we demonstrate that our method performs competitively against state-of-the-art OOD detection algorithms and also significantly boosts the semi-supervised learning performance for both ID and OOD classes.

Advanced AI models hold the promise of tremendous benefits for humanity, but society needs to proactively manage the accompanying risks. In this paper, we focus on what we term "frontier AI" models: highly capable foundation models that could possess dangerous capabilities sufficient to pose severe risks to public safety. Frontier AI models pose a distinct regulatory challenge: dangerous capabilities can arise unexpectedly; it is difficult to robustly prevent a deployed model from being misused; and, it is difficult to stop a model's capabilities from proliferating broadly. To address these challenges, at least three building blocks for the regulation of frontier models are needed: (1) standard-setting processes to identify appropriate requirements for frontier AI developers, (2) registration and reporting requirements to provide regulators with visibility into frontier AI development processes, and (3) mechanisms to ensure compliance with safety standards for the development and deployment of frontier AI models. Industry self-regulation is an important first step. However, wider societal discussions and government intervention will be needed to create standards and to ensure compliance with them. We consider several options to this end, including granting enforcement powers to supervisory authorities and licensure regimes for frontier AI models. Finally, we propose an initial set of safety standards. These include conducting pre-deployment risk assessments; external scrutiny of model behavior; using risk assessments to inform deployment decisions; and monitoring and responding to new information about model capabilities and uses post-deployment. We hope this discussion contributes to the broader conversation on how to balance public safety risks and innovation benefits from advances at the frontier of AI development., Comment: Update July 11th: - Added missing footnote back in. - Adjusted author order (mistakenly non-alphabetical among the first 6 authors) and adjusted affiliations (Jess Whittlestone's affiliation was mistagged and Gillian Hadfield had SRI added to her affiliations) Updated September 4th: Various typos

For $g\ge 2$ and $n\ge 0$, let $\mathcal{H}_{g,n}\subset \mathcal{M}_{g,n}$ denote the complex moduli stack of $n$-marked smooth hyperelliptic curves of genus $g$. A normal crossings compactification of this space is provided by the theory of pointed admissible $\mathbb{Z}/2\mathbb{Z}$-covers. We explicitly determine the resulting dual complex, and we use this to define a graph complex which computes the weight zero compactly supported cohomology of $\mathcal{H}_{g, n}$. Using this graph complex, we give a sum-over-graphs formula for the $S_n$-equivariant weight zero compactly supported Euler characteristic of $\mathcal{H}_{g, n}$. This formula allows for the computer-aided calculation, for each $g\le 7$, of the generating function $\mathsf{h}_g$ for these equivariant Euler characteristics for all $n$. More generally, we determine the dual complex of the boundary in any moduli space of pointed admissible $G$-covers of genus zero curves, when $G$ is abelian, as a symmetric $\Delta$-complex. We use these complexes to generalize our formula for $\mathsf{h}_g$ to moduli spaces of $n$-pointed smooth abelian covers of genus zero curves., Comment: Published version

To realize human-robot collaboration, robots need to execute actions for new tasks according to human instructions given finite prior knowledge. Human experts can share their knowledge of how to perform a task with a robot through multi-modal instructions in their demonstrations, showing a sequence of short-horizon steps to achieve a long-horizon goal. This paper introduces a method for robot action sequence generation from instruction videos using (1) an audio-visual Transformer that converts audio-visual features and instruction speech to a sequence of robot actions called dynamic movement primitives (DMPs) and (2) style-transfer-based training that employs multi-task learning with video captioning and weakly-supervised learning with a semantic classifier to exploit unpaired video-action data. We built a system that accomplishes various cooking actions, where an arm robot executes a DMP sequence acquired from a cooking video using the audio-visual Transformer. Experiments with Epic-Kitchen-100, YouCookII, QuerYD, and in-house instruction video datasets show that the proposed method improves the quality of DMP sequences by 2.3 times the METEOR score obtained with a baseline video-to-action Transformer. The model achieved 32% of the task success rate with the task knowledge of the object., Comment: Accepted to Interspeech2023

Although Lorentz symmetry is a staple of General Relativity (GR), there are several reasons to believe it may not hold in a more advanced theory of gravity, such as quantum gravity. Einstein-aether theory is a modified theory of gravity that breaks Lorentz symmetry by introducing a dynamical vector field called the aether. The theory has four coupling constants that characterize deviations from GR and that must be determined through observations. Although three of the four parameters have been constrained by various empirical observations and stability requirements, one, called $c_\omega$, remains essentially unconstrained. The aim of this work is to see if a constraint on $c_\omega$ can be derived from the I-Love-Q universal relations for neutron stars, which connect the neutron star moment of inertia (I), the tidal Love number (Love), and the quadrupole moment (Q) in a way that is insensitive to uncertainties in the neutron star equation-of-state. To understand if the theory can be constrained through such relations, we model slowly-rotating or weakly tidally-deformed neutron stars in Einstein-aether theory, derive their I-Love-Q relations, and study how they depend on $c_\omega$. We find that the I-Love-Q relations in Einstein-aether theory are insensitive to $c_\omega$ and that they are close to the relations in GR. This means that the I-Love-Q relations in Einstein-aether theory remain universal but cannot be used to constrain the theory. These results indicate that to constrain the theory with neutron stars, it is necessary to investigate relations involving other observables., Comment: 20 pages, 6 figures

Open-source reinforcement learning (RL) environments have played a crucial role in driving progress in the development of AI algorithms. In modern RL research, there is a need for simulated environments that are performant, scalable, and modular to enable their utilization in a wider range of potential real-world applications. Therefore, we present Jumanji, a suite of diverse RL environments specifically designed to be fast, flexible, and scalable. Jumanji provides a suite of environments focusing on combinatorial problems frequently encountered in industry, as well as challenging general decision-making tasks. By leveraging the efficiency of JAX and hardware accelerators like GPUs and TPUs, Jumanji enables rapid iteration of research ideas and large-scale experimentation, ultimately empowering more capable agents. Unlike existing RL environment suites, Jumanji is highly customizable, allowing users to tailor the initial state distribution and problem complexity to their needs. Furthermore, we provide actor-critic baselines for each environment, accompanied by preliminary findings on scaling and generalization scenarios. Jumanji aims to set a new standard for speed, adaptability, and scalability of RL environments., Comment: 9 pages + 21 pages of appendices and references. Published at ICLR 2024

The goal of offline black-box optimization (BBO) is to optimize an expensive black-box function using a fixed dataset of function evaluations. Prior works consider forward approaches that learn surrogates to the black-box function and inverse approaches that directly map function values to corresponding points in the input domain of the black-box function. These approaches are limited by the quality of the offline dataset and the difficulty in learning one-to-many mappings in high dimensions, respectively. We propose Denoising Diffusion Optimization Models (DDOM), a new inverse approach for offline black-box optimization based on diffusion models. Given an offline dataset, DDOM learns a conditional generative model over the domain of the black-box function conditioned on the function values. We investigate several design choices in DDOM, such as re-weighting the dataset to focus on high function values and the use of classifier-free guidance at test-time to enable generalization to function values that can even exceed the dataset maxima. Empirically, we conduct experiments on the Design-Bench benchmark and show that DDOM achieves results competitive with state-of-the-art baselines., Comment: International Conference on Machine Learning 2023

Prediction markets elicit and aggregate beliefs by paying agents based on how close their predictions are to a verifiable future outcome. However, outcomes of many important questions are difficult to verify or unverifiable, in that the ground truth may be hard or impossible to access. Examples include questions about causal effects where it is infeasible or unethical to run randomized trials; crowdsourcing and content moderation tasks where it is prohibitively expensive to verify ground truth; and questions asked over long time horizons, where the delay until the realization of the outcome skews agents' incentives to report their true beliefs. We present a novel and unintuitive result showing that it is possible to run an $\varepsilon-$incentive compatible prediction market to elicit and efficiently aggregate information from a pool of agents without observing the outcome by paying agents the negative cross-entropy between their prediction and that of a carefully chosen reference agent. Our key insight is that a reference agent with access to more information can serve as a reasonable proxy for the ground truth. We use this insight to propose self-resolving prediction markets that terminate with some probability after every report and pay all but a few agents based on the final prediction. We show that it is an $\varepsilon-$Perfect Bayesian Equilibrium for all agents to report truthfully in our mechanism and to believe that all other agents report truthfully. Although primarily of interest for unverifiable outcomes, this design is also applicable for verifiable outcomes.

Generative adversarial networks (GANs) comprise a generator, trained to learn the underlying distribution of the desired data, and a discriminator, trained to distinguish real samples from those output by the generator. A majority of GAN literature focuses on understanding the optimality of the discriminator through integral probability metric (IPM) or divergence based analysis. In this paper, we propose a unified approach to analyzing the generator optimization through variational approach. In $f$-divergence-minimizing GANs, we show that the optimal generator is the one that matches the score of its output distribution with that of the data distribution, while in IPM GANs, we show that this optimal generator matches score-like functions, involving the flow-field of the kernel associated with a chosen IPM constraint space. Further, the IPM-GAN optimization can be seen as one of smoothed score-matching, where the scores of the data and the generator distributions are convolved with the kernel associated with the constraint. The proposed approach serves to unify score-based training and existing GAN flavors, leveraging results from normalizing flows, while also providing explanations for empirical phenomena such as the stability of non-saturating GAN losses. Based on these results, we propose novel alternatives to $f$-GAN and IPM-GAN training based on score and flow matching, and discriminator-guided Langevin sampling.

We consider the problem of optimizing the discriminator in generative adversarial networks (GANs) subject to higher-order gradient regularization. We show analytically, via the least-squares (LSGAN) and Wasserstein (WGAN) GAN variants, that the discriminator optimization problem is one of interpolation in $n$-dimensions. The optimal discriminator, derived using variational Calculus, turns out to be the solution to a partial differential equation involving the iterated Laplacian or the polyharmonic operator. The solution is implementable in closed-form via polyharmonic radial basis function (RBF) interpolation. In view of the polyharmonic connection, we refer to the corresponding GANs as Poly-LSGAN and Poly-WGAN. Through experimental validation on multivariate Gaussians, we show that implementing the optimal RBF discriminator in closed-form, with penalty orders $m \approx\lceil \frac{n}{2} \rceil $, results in superior performance, compared to training GAN with arbitrarily chosen discriminator architectures. We employ the Poly-WGAN discriminator to model the latent space distribution of the data with encoder-decoder-based GAN flavors such as Wasserstein autoencoders.

Current approaches to building general-purpose AI systems tend to produce systems with both beneficial and harmful capabilities. Further progress in AI development could lead to capabilities that pose extreme risks, such as offensive cyber capabilities or strong manipulation skills. We explain why model evaluation is critical for addressing extreme risks. Developers must be able to identify dangerous capabilities (through "dangerous capability evaluations") and the propensity of models to apply their capabilities for harm (through "alignment evaluations"). These evaluations will become critical for keeping policymakers and other stakeholders informed, and for making responsible decisions about model training, deployment, and security., Comment: Fixed typos; added citation

The need for deep neural network (DNN) models with higher performance and better functionality leads to the proliferation of very large models. Model training, however, requires intensive computation time and energy. Memristor-based compute-in-memory (CIM) modules can perform vector-matrix multiplication (VMM) in situ and in parallel, and have shown great promises in DNN inference applications. However, CIM-based model training faces challenges due to non-linear weight updates, device variations, and low-precision in analog computing circuits. In this work, we experimentally implement a mixed-precision training scheme to mitigate these effects using a bulk-switching memristor CIM module. Lowprecision CIM modules are used to accelerate the expensive VMM operations, with high precision weight updates accumulated in digital units. Memristor devices are only changed when the accumulated weight update value exceeds a pre-defined threshold. The proposed scheme is implemented with a system-on-chip (SoC) of fully integrated analog CIM modules and digital sub-systems, showing fast convergence of LeNet training to 97.73%. The efficacy of training larger models is evaluated using realistic hardware parameters and shows that that analog CIM modules can enable efficient mix-precision DNN training with accuracy comparable to full-precision software trained models. Additionally, models trained on chip are inherently robust to hardware variations, allowing direct mapping to CIM inference chips without additional re-training.

Training Generative adversarial networks (GANs) stably is a challenging task. The generator in GANs transform noise vectors, typically Gaussian distributed, into realistic data such as images. In this paper, we propose a novel approach for training GANs with images as inputs, but without enforcing any pairwise constraints. The intuition is that images are more structured than noise, which the generator can leverage to learn a more robust transformation. The process can be made efficient by identifying closely related datasets, or a ``friendly neighborhood'' of the target distribution, inspiring the moniker, Spider GAN. To define friendly neighborhoods leveraging proximity between datasets, we propose a new measure called the signed inception distance (SID), inspired by the polyharmonic kernel. We show that the Spider GAN formulation results in faster convergence, as the generator can discover correspondence even between seemingly unrelated datasets, for instance, between Tiny-ImageNet and CelebA faces. Further, we demonstrate cascading Spider GAN, where the output distribution from a pre-trained GAN generator is used as the input to the subsequent network. Effectively, transporting one distribution to another in a cascaded fashion until the target is learnt -- a new flavor of transfer learning. We demonstrate the efficacy of the Spider approach on DCGAN, conditional GAN, PGGAN, StyleGAN2 and StyleGAN3. The proposed approach achieves state-of-the-art Frechet inception distance (FID) values, with one-fifth of the training iterations, in comparison to their baseline counterparts on high-resolution small datasets such as MetFaces, Ukiyo-E Faces and AFHQ-Cats., Comment: CVPR 2023

The skill of pivoting an object with a robotic system is challenging for the external forces that act on the system, mainly given by contact interaction. The complexity increases when the same skills are required to generalize across different objects. This paper proposes a framework for learning robust and generalizable pivoting skills, which consists of three steps. First, we learn a pivoting policy on an ``unitary'' object using Reinforcement Learning (RL). Then, we obtain the object's feature space by supervised learning to encode the kinematic properties of arbitrary objects. Finally, to adapt the unitary policy to multiple objects, we learn data-driven projections based on the object features to adjust the state and action space of the new pivoting task. The proposed approach is entirely trained in simulation. It requires only one depth image of the object and can zero-shot transfer to real-world objects. We demonstrate robustness to sim-to-real transfer and generalization to multiple objects., Comment: 2023 International Conference on Robotics and Automation (ICRA)

In the field of cooperative multi-agent reinforcement learning (MARL), the standard paradigm is the use of centralised training and decentralised execution where a central critic conditions the policies of the cooperative agents based on a central state. It has been shown, that in cases with large numbers of redundant agents these methods become less effective. In a more general case, there is likely to be a larger number of agents in an environment than is required to solve the task. These redundant agents reduce performance by enlarging the dimensionality of both the state space and and increasing the size of the joint policy used to solve the environment. We propose leveraging layerwise relevance propagation (LRP) to instead separate the learning of the joint value function and generation of local reward signals and create a new MARL algorithm: relevance decomposition network (RDN). We find that although the performance of both baselines VDN and Qmix degrades with the number of redundant agents, RDN is unaffected., Comment: Published at the 22nd International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2023). 2 Pages, 1 Figure

Numerous parties are calling for the democratisation of AI, but the phrase is used to refer to a variety of goals, the pursuit of which sometimes conflict. This paper identifies four kinds of AI democratisation that are commonly discussed: (1) the democratisation of AI use, (2) the democratisation of AI development, (3) the democratisation of AI profits, and (4) the democratisation of AI governance. Numerous goals and methods of achieving each form of democratisation are discussed. The main takeaway from this paper is that AI democratisation is a multifarious and sometimes conflicting concept that should not be conflated with improving AI accessibility. If we want to move beyond ambiguous commitments to democratising AI, to productive discussions of concrete policies and trade-offs, then we need to recognise the principal role of the democratisation of AI governance in navigating tradeoffs and risks across decisions around use, development, and profits., Comment: V2 Changed second author affiliation; added citation to section 5.2; edit to author contribution statement; V3 camera ready version for conference proceedings. Minor content changes in response to reviewer comments

Many generative foundation models (or GFMs) are trained on publicly available data and use public infrastructure, but 1) may degrade the "digital commons" that they depend on, and 2) do not have processes in place to return value captured to data producers and stakeholders. Existing conceptions of data rights and protection (focusing largely on individually-owned data and associated privacy concerns) and copyright or licensing-based models offer some instructive priors, but are ill-suited for the issues that may arise from models trained on commons-based data. We outline the risks posed by GFMs and why they are relevant to the digital commons, and propose numerous governance-based solutions that include investments in standardized dataset/model disclosure and other kinds of transparency when it comes to generative models' training and capabilities, consortia-based funding for monitoring/standards/auditing organizations, requirements or norms for GFM companies to contribute high quality data to the commons, and structures for shared ownership based on individual or community provision of fine-tuning data.

For energy efficient operation of the massive multiple-input multiple-output (MIMO) networks, various aspects of energy efficiency maximization have been addressed, where a careful selection of number of active antennas has shown significant gains. Moreover, switching-off physical resource blocks (PRBs) and carrier shutdown saves energy in low load scenarios. However, the joint optimization of spectral PRB allocation and spatial layering in a heterogeneous network has not been completely solved yet. Therefore, we study a power consumption model for multi-cell multi-user massive MIMO 5G network, capturing the joint effects of both dimensions. We characterize the optimal resource allocation under practical constraints, i.e., limited number of available antennas, PRBs, base stations (BSs), and frequency bands. We observe a single spatial layer achieving lowest energy consumption in very low load scenarios, whereas, spatial layering is required in high load scenarios. Finally, we derive novel algorithms for energy efficient user to BS assignment and propose an adaptive algorithm for PRB assignment and power control. All results are illustrated by numerical system-level simulations, describing a realistic metropolis scenario. The results show that a higher frequency band should be used to support users with large rate requirements via spatial multiplexing and assigning each user maximum available PRBs.

Representing and reasoning about uncertainty is crucial for autonomous agents acting in partially observable environments with noisy sensors. Partially observable Markov decision processes (POMDPs) serve as a general framework for representing problems in which uncertainty is an important factor. Online sample-based POMDP methods have emerged as efficient approaches to solving large POMDPs and have been shown to extend to continuous domains. However, these solutions struggle to find long-horizon plans in problems with significant uncertainty. Exploration heuristics can help guide planning, but many real-world settings contain significant task-irrelevant uncertainty that might distract from the task objective. In this paper, we propose STRUG, an online POMDP solver capable of handling domains that require long-horizon planning with significant task-relevant and task-irrelevant uncertainty. We demonstrate our solution on several temporally extended versions of toy POMDP problems as well as robotic manipulation of articulated objects using a neural perception frontend to construct a distribution of possible models. Our results show that STRUG outperforms the current sample-based online POMDP solvers on several tasks.

Robots have been steadily increasing their presence in our daily lives, where they can work along with humans to provide assistance in various tasks on industry floors, in offices, and in homes. Automated assembly is one of the key applications of robots, and the next generation assembly systems could become much more efficient by creating collaborative human-robot systems. However, although collaborative robots have been around for decades, their application in truly collaborative systems has been limited. This is because a truly collaborative human-robot system needs to adjust its operation with respect to the uncertainty and imprecision in human actions, ensure safety during interaction, etc. In this paper, we present a system for human-robot collaborative assembly using learning from demonstration and pose estimation, so that the robot can adapt to the uncertainty caused by the operation of humans. Learning from demonstration is used to generate motion trajectories for the robot based on the pose estimate of different goal locations from a deep learning-based vision system. The proposed system is demonstrated using a physical 6 DoF manipulator in a collaborative human-robot assembly scenario. We show successful generalization of the system's operation to changes in the initial and final goal locations through various experiments.

Satellite based Quantum Key Distribution (QKD) in Low Earth Orbit (LEO) is currently the only viable technology to span thousands of kilometres. Since the typical overhead pass of a satellite lasts for a few minutes, it is crucial to increase the the signal rate to maximise the secret key length. For the QUARC CubeSat mission due to be launched within two years, we are designing a dual wavelength, weak-coherent-pulse decoy-state Bennett-Brassard '84 (WCP DS BB84) QKD source. The optical payload is designed in a $12{\times}9{\times}5 cm^3$ bespoke aluminium casing. The Discrete Variable QKD Source consists of two symmetric sources operating at 785 nm and 808 nm. The laser diodes are fixed to produce Horizontal,Vertical, Diagonal, and Anti-diagonal (H,V,D,A) polarisation respectively, which are combined and attenuated to a mean photon number of 0.3 and 0.5 photons/pulse. We ensure that the source is secure against most side channel attacks by spatially mode filtering the output beam and characterising their spectral and temporal characterstics. The extinction ratio of the source contributes to the intrinsic Qubit Error Rate(QBER) with $0.817 \pm 0.001\%$. This source operates at 200MHz, which is enough to provide secure key rates of a few kilo bits per second despite 40 dB of estimated loss in the free space channel

Quantum networks have been shown to connect users with full-mesh topologies without trusted nodes. We present advancements on our scalable polarisation entanglement-based quantum network testbed, which has the ability to perform protocols beyond simple quantum key distribution. Our approach utilises wavelength multiplexing, which is ideal for quantum networks across local metropolitan areas due to the ease of connecting additional users to the network without increasing the resource requirements per user. We show a 10 user fully connected quantum network with metropolitan scale deployed fibre links, demonstrating polarisation stability and the ability to generate secret keys over a period of 10.8 days with a network wide average-effective secret key rate of 3.38 bps., Comment: 8 pages, 4 figures, 2 tables, SPIE Photonex 2022 conference proceedings

In this work, we investigate the effectiveness of pretrained Self-Supervised Learning (SSL) features for learning the mapping for acoustic to articulatory inversion (AAI). Signal processing-based acoustic features such as MFCCs have been predominantly used for the AAI task with deep neural networks. With SSL features working well for various other speech tasks such as speech recognition, emotion classification, etc., we experiment with its efficacy for AAI. We train on SSL features with transformer neural networks-based AAI models of 3 different model complexities and compare its performance with MFCCs in subject-specific (SS), pooled and fine-tuned (FT) configurations with data from 10 subjects, and evaluate with correlation coefficient (CC) score on the unseen sentence test set. We find that acoustic feature reconstruction objective-based SSL features such as TERA and DeCoAR work well for AAI, with SS CCs of these SSL features reaching close to the best FT CCs of MFCC. We also find the results consistent across different model sizes., Comment: submitted to ICASSP 2023

We formulate and prove a version of the arithmetic Siegel-Weil formula for (zero dimensional) Shimura varieties attached to tori, equipped with some additional data. More precisely, we define a family of ``special" divisors in terms of Green functions at archimedean and non-archimedean places, and prove that their degrees coincide with the Fourier coefficients of the central derivative of an Eisenstein series. The proof relies on the usual Siegel-Weil formula to provide a direct link between both sides of the identity, and in some sense, offers a more conceptual point of view on prior results in the literature.

Measurement error mitigation (MEM) techniques are postprocessing strategies to counteract systematic read-out errors on quantum computers (QC). Currently used MEM strategies face a tradeoff: methods that scale well with the number of qubits return negative probabilities, while those that guarantee a valid probability distribution are not scalable. Here, we present a scheme that addresses both of these issues. In particular, we present a scalable implementation of iterative Bayesian unfolding, a standard mitigation technique used in high-energy physics experiments. We demonstrate our method by mitigating QC data from experimental preparation of Greenberger-Horne-Zeilinger (GHZ) states up to 127 qubits and implementation of the Bernstein-Vazirani algorithm on up to 26 qubits.

Quantum key distribution (QKD) is a theoretically proven future-proof secure encryption method that inherits its security from fundamental physical principles. Craft Prospect, working with a number of UK organisations, has been focused on miniaturising the technologies that enable QKD so that they may be used in smaller platforms including nanosatellites. The significant reduction of size, and therefore the cost of launching quantum communication technologies either on a dedicated platform or hosted as part of a larger optical communications will improve potential access to quantum encryption on a relatively quick timescale. The ROKS mission seeks to be among the first to send a QKD payload on a CubeSat into low Earth orbit, demonstrating the capabilities of newly developed modular quantum technologies. The ROKS payload comprises a quantum source module that supplies photons randomly in any of four linear polarisation states fed from a quantum random number generator; an acquisition, pointing, and tracking system to fine-tune alignment of the quantum source beam with an optical ground station; an imager that will detect cloud cover autonomously; and an onboard computer that controls and monitors the other modules, which manages the payload and assures the overall performance and security of the system. Each of these modules have been developed with low SWaP for CubeSats, but with interoperability in mind for other satellite form factors. We present each of the listed components, together with the initial test results from our test bench and the performance of our protoflight models prior to initial integration with the 6U CubeSat platform systems. The completed ROKS payload will be ready for flight at the end of 2022, with various modular components already being baselined for flight and integrated into third party communication missions., Comment: 13 pages with 25 figures. Presented at Small Satellite Conference: https://digitalcommons.usu.edu/smallsat/2022/all2022/163/. Any comments are welcome

We demonstrate a novel MIR methane sensor shifting measurement wavelength to SWIR (1.55$\mu$m) by using non-linear interferometry. The technique exploits the interference effects seen in three-wave mixing when pump, signal, and idler modes make a double pass through a nonlinear crystal. The method allows sensing at wavelengths where detectors are poor ($>$3$\mu$m) and detection at wavelengths where photon counting sensitivity can be achieved. In a first experimental demonstration, we measured a small methane concentration inside a gas cell with high precision. This interferometer can be built in a compact design for field operations and potentially enable the detection of low concentrations of methane at up to 100m range. Signal-to-noise ratio calculations show that the method can outperform existing short wavelength ($\sim$1.65$\mu$m) integrated path differential absorption direct sensing at high ($>$$10^{-4}$) non-linear gain., Comment: Title change, new data, significant revision, 10 pages 5 figures

Multi-agent reinforcement learning (MARL) has emerged as a useful approach to solving decentralised decision-making problems at scale. Research in the field has been growing steadily with many breakthrough algorithms proposed in recent years. In this work, we take a closer look at this rapid development with a focus on evaluation methodologies employed across a large body of research in cooperative MARL. By conducting a detailed meta-analysis of prior work, spanning 75 papers accepted for publication from 2016 to 2022, we bring to light worrying trends that put into question the true rate of progress. We further consider these trends in a wider context and take inspiration from single-agent RL literature on similar issues with recommendations that remain applicable to MARL. Combining these recommendations, with novel insights from our analysis, we propose a standardised performance evaluation protocol for cooperative MARL. We argue that such a standard protocol, if widely adopted, would greatly improve the validity and credibility of future research, make replication and reproducibility easier, as well as improve the ability of the field to accurately gauge the rate of progress over time by being able to make sound comparisons across different works. Finally, we release our meta-analysis data publicly on our project website for future research on evaluation: https://sites.google.com/view/marl-standard-protocol, Comment: Published at the 36th Conference on Neural Information Processing Systems (NeurIPS 2022). Website: see https://sites.google.com/view/marl-standard-protocol . 43 Pages, 21 Figures, 8 Tables

Over the last few decades, many aspects of human life have been enhanced with virtual domains, from the advent of digital assistants such as Amazon's Alexa and Apple's Siri to the latest metaverse efforts of the rebranded Meta. These trends underscore the importance of generating photorealistic visual depictions of humans. This has led to the rapid growth of so-called deepfake and talking-head generation methods in recent years. Despite their impressive results and popularity, they usually lack certain qualitative aspects such as texture quality, lips synchronization, or resolution, and practical aspects such as the ability to run in real-time. To allow for virtual human avatars to be used in practical scenarios, we propose an end-to-end framework for synthesizing high-quality virtual human faces capable of speaking with accurate lip motion with a special emphasis on performance. We introduce a novel network utilizing visemes as an intermediate audio representation and a novel data augmentation strategy employing a hierarchical image synthesis approach that allows disentanglement of the different modalities used to control the global head motion. Our method runs in real-time, and is able to deliver superior results compared to the current state-of-the-art.

The information-theoretic unconditional security offered by quantum key distribution has spurred the development of larger quantum communication networks. However, as these networks grow so does the strong need to reduce complexity and overheads. Polarization based entanglement distribution networks are a promising approach due to their scalability and lack of trusted nodes. Nevertheless, they are only viable if the birefringence of all optical distribution fibres in the network is compensated to preserve the polarization based quantum state. The brute force approach would require a few hundred fibre polarization controllers for even a moderately sized network. Instead, we propose and investigate four different methods of polarization compensation. We compare them based on complexity, effort, level of disruption to network operations and performance.

GPU technology has been improving at an expedited pace in terms of size and performance, empowering HPC and AI/ML researchers to advance the scientific discovery process. However, this also leads to inefficient resource usage, as most GPU workloads, including complicated AI/ML models, are not able to utilize the GPU resources to their fullest extent -- encouraging support for GPU multi-tenancy. We propose MISO, a technique to exploit the Multi-Instance GPU (MIG) capability on the latest NVIDIA datacenter GPUs (e.g., A100, H100) to dynamically partition GPU resources among co-located jobs. MISO's key insight is to use the lightweight, more flexible Multi-Process Service (MPS) capability to predict the best MIG partition allocation for different jobs, without incurring the overhead of implementing them during exploration. Due to its ability to utilize GPU resources more efficiently, MISO achieves 49% and 16% lower average job completion time than the unpartitioned and optimal static GPU partition schemes, respectively.

Streamline tracing in conical hypersonic flows is essential for designing high-performance waverider and intake. Conventionally, the streamline equations are solved after obtaining the velocity field from the solution of the axisymmetric conical flow field. The hypersonic waverider shape is generated from the base conical flow field by repeatedly applying the streamline tracing approach along several planes. When exploring the design space for optimization of the waverider, streamline tracing can be computationally expensive. We provide a novel strategy where first the Taylor-Maccoll equations for the inviscid axisymmetric conical flowfield and the streamlines from the shock are solved for a wide range of cone angle and Mach number conditions resulting in an extensive database. The streamlines are parametrized by a third-order polynomial, and an Artificial Neural Network (ANN) is trained to predict the coefficients of the polynomial for arbitrary inputs of Mach number, cone angle, and streamline originating location on the shock . We apply this strategy to design a cone derived waverider and compare the geometry obtained with the standard conical waverider design method and the simplified waverider design method. The ANN technique is highly accurate, with a difference of 0.68% with the standard in the coordinates of the waverider. RANS computations show that the ANN derived waverider does not indicate severe flow spillage at the leading edge, which is observed in the waverider generated from the simplified method. The new ANN-based approach is 20 times faster than the conventional method.

Ho\v{r}ava-Lifshitz gravity is an alternative theory to general relativity which breaks Lorentz invariance in order to achieve an ultraviolet complete and power-counting renormalizable theory of gravity. In the low energy limit, Ho\v{r}ava-Lifshitz gravity coincides with a vector-tensor theory known as khronometric gravity. The deviation of khronometric gravity from general relativity can be parametrized by three coupling constants: $\alpha$, $\beta$, and $\lambda$. Solar system experiments and gravitational wave observations impose stringent bounds on $\alpha$ and $\beta$, while $\lambda$ is still relatively unconstrained ($\lambda\lesssim 0.01$). In this paper, we study whether one can constrain this remaining parameter with neutron star observations through the universal I-Love-Q relations between the moment of inertia (I), the tidal Love number (Love), and the quadrupole moment (Q), which are insensitive to details in the nuclear matter equation of state. To do so, we perturbatively construct slowly-rotating and weakly tidally-deformed neutron stars in khronometric gravity. We find that the I-Love-Q relations are independent of $\lambda$ in the limit $(\alpha,\beta) \to 0$. Although some components of the field equations depend on $\lambda$, we show through induction and a post-Minkowskian analysis that slowly-rotating neutron stars do not depend on $\lambda$ at all. Tidally deformed neutron stars, on the other hand, are modified in khronometric gravity (though the usual Love number is not modified, as mentioned earlier), and there are potentially new, non-GR Love numbers, though their observability is unclear. These findings indicate that it may be difficult to constrain $\lambda$ with rotating/tidally-deformed neutron stars., Comment: 18 pages. Added a new Appendix, added additional explanation to Sects. II and III, and added a small clarification to Sect. V, no change to main results

Let $\bar{\mathcal{M}}_{g, m|n}$ denote Hassett's moduli space of weighted pointed stable curves of genus $g$ for the heavy/light weight data $\left(1^{(m)}, 1/n^{(n)}\right)$, and let $\mathcal{M}_{g, m|n} \subset \bar{\mathcal{M}}_{g, m|n}$ be the locus parameterizing smooth, not necessarily distinctly marked curves. We give a change-of-variables formula which computes the generating function for $(S_m\times S_n)$-equivariant Hodge-Deligne polynomials of these spaces in terms of the generating functions for $S_{n}$-equivariant Hodge-Deligne polynomials of $\bar{\mathcal{M}}_{g,n}$ and $\mathcal{M}_{g,n}$., Comment: 21 pages, 3 tables. Edits based on referee suggestions

Many problems in science and engineering involve optimizing an expensive black-box function over a high-dimensional space. For such black-box optimization (BBO) problems, we typically assume a small budget for online function evaluations, but also often have access to a fixed, offline dataset for pretraining. Prior approaches seek to utilize the offline data to approximate the function or its inverse but are not sufficiently accurate far from the data distribution. We propose BONET, a generative framework for pretraining a novel black-box optimizer using offline datasets. In BONET, we train an autoregressive model on fixed-length trajectories derived from an offline dataset. We design a sampling strategy to synthesize trajectories from offline data using a simple heuristic of rolling out monotonic transitions from low-fidelity to high-fidelity samples. Empirically, we instantiate BONET using a causally masked Transformer and evaluate it on Design-Bench, where we rank the best on average, outperforming state-of-the-art baselines., Comment: International Conference for Machine Learning 2023 NeurIPS Workshop for Foundational Models for Decision Making (Oral) 2022

We define a family of arithmetic zero cycles in the arithmetic Chow group of a modular curve X_0(N), for N>3 odd and squarefree, and identify the arithmetic degrees of these cycles as q-coefficients of the central derivative of a Siegel Eisenstein series of genus two. This parallels work of Kudla-Rapoport-Yang for Shimura curves., Comment: 43 pages