Search

Systems equipped with modern sensing modalities such as vision and lidar gain access to increasingly high-dimensional measurements with which to enact estimation and control schemes. In this article, we examine the continuum limit of high-dimensional measurements and analyze state estimation in linear time-invariant systems with infinite-dimensional measurements but finite-dimensional states, both corrupted by additive noise. We propose a linear filter and derive the corresponding optimal gain functional in the sense of the minimum mean square error, analogous to the classic Kalman filter. By modeling the measurement noise as a wide-sense stationary random field, we are able to derive the optimal linear filter explicitly, in contrast to previous derivations of Kalman filters in distributed-parameter settings. Interestingly, we find that we need only impose conditions that are finite-dimensional in nature to ensure that the filter is asymptotically stable. The proposed filter is verified via simulation of a linearized system with a pinhole camera sensor., Comment: 15 pages, 3 Figures. This paper will appear in the IEEE Transactions on Automatic Control

Automotive simulation can potentially compensate for a lack of training data in computer vision applications. However, there has been little to no image quality evaluation of automotive simulation and the impact of optical degradations on simulation is little explored. In this work, we investigate Virtual KITTI and the impact of applying variations of Gaussian blur on image sharpness. Furthermore, we consider object detection, a common computer vision application on three different state-of-the-art models, thus allowing us to characterize the relationship between object detection and sharpness. It was found that while image sharpness (MTF50) degrades from an average of 0.245cy/px to approximately 0.119cy/px; object detection performance stays largely robust within 0.58\%(Faster RCNN), 1.45\%(YOLOF) and 1.93\%(DETR) across all respective held-out test sets., Comment: 8 pages, 2 figures, 2 tables. This paper is a preprint of a paper submitted to the 26th Irish Machine Vision and Image Processing Conference (IMVIP 2024). If accepted, the copy of record will be available at IET Digital Library

We formulate the discrete-time inverse optimal control problem of inferring unknown parameters in the objective function of an optimal control problem from measurements of optimal states and controls as a nonlinear filtering problem. This formulation enables us to propose a novel extended Kalman filter (EKF) for solving inverse optimal control problems in a computationally efficient recursive online manner that requires only a single pass through the measurement data. Importantly, we show that the Jacobians required to implement our EKF can be computed efficiently by exploiting recent Pontryagin differentiable programming results, and that our consideration of an EKF enables the development of first-of-their-kind theoretical error guarantees for online inverse optimal control with noisy incomplete measurements. Our proposed EKF is shown to be significantly faster than an alternative unscented Kalman filter-based approach., Comment: 7 pages, 2 figures, accepted for presentation at 2024 American Control Conference

In this report, we introduce the Gemini 1.5 family of models, representing the next generation of highly compute-efficient multimodal models capable of recalling and reasoning over fine-grained information from millions of tokens of context, including multiple long documents and hours of video and audio. The family includes two new models: (1) an updated Gemini 1.5 Pro, which exceeds the February version on the great majority of capabilities and benchmarks; (2) Gemini 1.5 Flash, a more lightweight variant designed for efficiency with minimal regression in quality. Gemini 1.5 models achieve near-perfect recall on long-context retrieval tasks across modalities, improve the state-of-the-art in long-document QA, long-video QA and long-context ASR, and match or surpass Gemini 1.0 Ultra's state-of-the-art performance across a broad set of benchmarks. Studying the limits of Gemini 1.5's long-context ability, we find continued improvement in next-token prediction and near-perfect retrieval (>99%) up to at least 10M tokens, a generational leap over existing models such as Claude 3.0 (200k) and GPT-4 Turbo (128k). Finally, we highlight real-world use cases, such as Gemini 1.5 collaborating with professionals on completing their tasks achieving 26 to 75% time savings across 10 different job categories, as well as surprising new capabilities of large language models at the frontier; when given a grammar manual for Kalamang, a language with fewer than 200 speakers worldwide, the model learns to translate English to Kalamang at a similar level to a person who learned from the same content.

It is becoming increasingly important in the software industry, especially with the growing complexity of software ecosystems and the emphasis on security and compliance for manufacturers to inventory software used on their systems. A Software-Bill-of-Materials (SBOM) is a comprehensive inventory detailing a software application's components and dependencies. Current approaches rely on case-based reasoning to inconsistently identify the software components embedded in binary files. We propose a different route, an automated method for generating SBOMs to prevent disastrous supply-chain attacks. Remaining on the topic of static code analysis, we interpret this problem as a semantic similarity task wherein a transformer model can be trained to relate a product name to corresponding version strings. Our test results are compelling, demonstrating the model's strong performance in the zero-shot classification task, further demonstrating the potential for use in a real-world cybersecurity context., Comment: 8 pages, 2 figures

Training of modern large neural networks (NN) requires a combination of parallelization strategies encompassing data, model, or optimizer sharding. When strategies increase in complexity, it becomes necessary for partitioning tools to be 1) expressive, allowing the composition of simpler strategies, and 2) predictable to estimate performance analytically. We present PartIR, our design for a NN partitioning system. PartIR is focused on an incremental approach to rewriting and is hardware-and-runtime agnostic. We present a simple but powerful API for composing sharding strategies and a simulator to validate them. The process is driven by high-level programmer-issued partitioning tactics, which can be both manual and automatic. Importantly, the tactics are specified separately from the model code, making them easy to change. We evaluate PartIR on several different models to demonstrate its predictability, expressibility, and ability to reach peak performance..

The Latent Block Model (LBM) is a prominent model-based co-clustering method, returning parametric representations of each block cluster and allowing the use of well-grounded model selection methods. The LBM, while adapted in literature to handle different feature types, cannot be applied to datasets consisting of multiple disjoint sets of features, termed views, for a common set of observations. In this work, we introduce the multi-view LBM, extending the LBM method to multi-view data, where each view marginally follows an LBM. In the case of two views, the dependence between them is captured by a cluster membership matrix, and we aim to learn the structure of this matrix. We develop a likelihood-based approach in which parameter estimation uses a stochastic EM algorithm integrating a Gibbs sampler, and an ICL criterion is derived to determine the number of row and column clusters in each view. To motivate the application of multi-view methods, we extend recent work developing hypothesis tests for the null hypothesis that clusters of observations in each view are independent of each other. The testing procedure is integrated into the model estimation strategy. Furthermore, we introduce a penalty scheme to generate sparse row clusterings. We verify the performance of the developed algorithm using synthetic datasets, and provide guidance for optimal parameter selection. Finally, the multi-view co-clustering method is applied to a complex genomics dataset, and is shown to provide new insights for high-dimension multi-view problems.

We study the 2-offer semirandom 3-uniform hypergraph model on $n$ vertices. At each step, we are presented with 2 uniformly random vertices. We choose any other vertex, thus creating a hyperedge of size 3. We show a strategy that constructs a perfect matching, and another that constructs a loose Hamilton cycle, both succeeding asymptotically almost surely within $\Theta(n)$ steps. Both results extend to $s$-uniform hypergraphs. The challenges with hypergraphs, and our methods, are qualitatively different from what has been seen for semirandom graphs. Much of our analysis is done on an auxiliary graph that is a uniform $k$-out subgraph of a random bipartite graph, and this tool may be useful in other contexts., Comment: This version makes the title more explicit, revises the introduction and background, corrects some minor errors, and adds a few clarifications

This report introduces a new family of multimodal models, Gemini, that exhibit remarkable capabilities across image, audio, video, and text understanding. The Gemini family consists of Ultra, Pro, and Nano sizes, suitable for applications ranging from complex reasoning tasks to on-device memory-constrained use-cases. Evaluation on a broad range of benchmarks shows that our most-capable Gemini Ultra model advances the state of the art in 30 of 32 of these benchmarks - notably being the first model to achieve human-expert performance on the well-studied exam benchmark MMLU, and improving the state of the art in every one of the 20 multimodal benchmarks we examined. We believe that the new capabilities of the Gemini family in cross-modal reasoning and language understanding will enable a wide variety of use cases. We discuss our approach toward post-training and deploying Gemini models responsibly to users through services including Gemini, Gemini Advanced, Google AI Studio, and Cloud Vertex AI.

Laser technologies improved after the understanding of the Chirped Pulse Amplification (CPA) which allows energetic laser beams to be compressed to tens of femtosecond (fs) pulse durations and focused to few $\mu$m. Protons of tens of MeV can be accelerated using for instance the Target Normal Sheath Acceleration (TNSA) method and focused on secondary targets. In such conditions, nuclear reactions can occur and radioisotopes relevant for medical purposes be produced. High repetition lasers can be used to produce enough isotopes for medical applications. This route is competitive to conventional methods mostly based on accelerators. In this paper we study the production of $^{67}$Cu, $^{63}$Zn, $^{18}$F and $^{11}$C currently used in positron emission tomography (PET) and other applications. At the same time, we study the reaction $^{10}$B(p,$\alpha$)$^{7}$Be and $^{70}$Zn(p,4n)$^{67}$Ga to put further constraints to the proton distributions at different angles and to the reaction $^{11}$B(p,$\alpha$)$^{8}$Be relevant for energy production. The experiment was performed at the 1 petawatt (PW) laser facility at Vega III located in Salamanca-Spain. Angular distributions of radioisotopes in the forward (with respect to the laser direction) and backward directions were measured using a High Purity Germanium Detector (HPGE). Our results are reasonably reproduced by the numerical estimates following the approach of Kimura et al. (NIMA637(2011)167)

Assurance Cases (ACs) are an established approach in safety engineering to argue quality claims in a structured way. In the context of quality assurance for Machine Learning (ML)-based software components, ACs are also being discussed and appear promising. Tools for operationalizing ACs do exist, yet mainly focus on supporting safety engineers on the system level. However, assuring the quality of an ML component within the system is commonly the responsibility of data scientists, who are usually less familiar with these tools. To address this gap, we propose a framework to support the operationalization of ACs for ML components based on technologies that data scientists use on a daily basis: Python and Jupyter Notebook. Our aim is to make the process of creating ML-related evidence in ACs more effective. Results from the application of the framework, documented through notebooks, can be integrated into existing AC tools. We illustrate the application of the framework on an example excerpt concerned with the quality of the test data., Comment: Accepted for publication at International Conference on Product-Focused Software Process Improvement (Profes 2023), https://conf.researchr.org/home/profes-2023

This paper proposes a new method for differentiating through optimal trajectories arising from non-convex, constrained discrete-time optimal control (COC) problems using the implicit function theorem (IFT). Previous works solve a differential Karush-Kuhn-Tucker (KKT) system for the trajectory derivative, and achieve this efficiently by solving an auxiliary Linear Quadratic Regulator (LQR) problem. In contrast, we directly evaluate the matrix equations which arise from applying variable elimination on the Lagrange multiplier terms in the (differential) KKT system. By appropriately accounting for the structure of the terms within the resulting equations, we show that the trajectory derivatives scale linearly with the number of timesteps. Furthermore, our approach allows for easy parallelization, significantly improved scalability with model size, direct computation of vector-Jacobian products and improved numerical stability compared to prior works. As an additional contribution, we unify prior works, addressing claims that computing trajectory derivatives using IFT scales quadratically with the number of timesteps. We evaluate our method on a both synthetic benchmark and four challenging, learning from demonstration benchmarks including a 6-DoF maneuvering quadrotor and 6-DoF rocket powered landing., Comment: Accepted to NeurIPS 2023 (poster)

The second data release of the LOFAR Two-Metre Sky Survey (LoTSS) covers 27% of the northern sky, with a total area of $\sim 5,700$ deg$^2$. The high angular resolution of LOFAR with Dutch baselines (6 arcsec) allows us to carry out optical identifications of a large fraction of the detected radio sources without further radio followup; however, the process is made more challenging by the many extended radio sources found in LOFAR images as a result of its excellent sensitivity to extended structure. In this paper we present source associations and identifications for sources in the second data release based on optical and near-infrared data, using a combination of a likelihood-ratio cross-match method developed for our first data release, our citizen science project Radio Galaxy Zoo: LOFAR, and new approaches to algorithmic optical identification, together with extensive visual inspection by astronomers. We also present spectroscopic or photometric redshifts for a large fraction of the optical identifications. In total 4,116,934 radio sources lie in the area with good optical data, of which 85% have an optical or infrared identification and 58% have a good redshift estimate. We demonstrate the quality of the dataset by comparing it with earlier optically identified radio surveys. This is by far the largest ever optically identified radio catalogue, and will permit robust statistical studies of star-forming and radio-loud active galaxies., Comment: 29 pages. Accepted by A&A; data products available at https://lofar-surveys.org/dr2_release.html

Machine learning models are known to be vulnerable to adversarial evasion attacks as illustrated by image classification models. Thoroughly understanding such attacks is critical in order to ensure the safety and robustness of critical AI tasks. However, most evasion attacks are difficult to deploy against a majority of AI systems because they have focused on image domain with only few constraints. An image is composed of homogeneous, numerical, continuous, and independent features, unlike many other input types to AI systems used in practice. Furthermore, some input types include additional semantic and functional constraints that must be observed to generate realistic adversarial inputs. In this work, we propose a new framework to enable the generation of adversarial inputs irrespective of the input type and task domain. Given an input and a set of pre-defined input transformations, our framework discovers a sequence of transformations that result in a semantically correct and functional adversarial input. We demonstrate the generality of our approach on several diverse machine learning tasks with various input representations. We also show the importance of generating adversarial examples as they enable the deployment of mitigation techniques., Comment: Accepted at USENIX '23

Contrastive pretrained large Vision-Language Models (VLMs) like CLIP have revolutionized visual representation learning by providing good performance on downstream datasets. VLMs are 0-shot adapted to a downstream dataset by designing prompts that are relevant to the dataset. Such prompt engineering makes use of domain expertise and a validation dataset. Meanwhile, recent developments in generative pretrained models like GPT-4 mean they can be used as advanced internet search tools. They can also be manipulated to provide visual information in any structure. In this work, we show that GPT-4 can be used to generate text that is visually descriptive and how this can be used to adapt CLIP to downstream tasks. We show considerable improvements in 0-shot transfer accuracy on specialized fine-grained datasets like EuroSAT (~7%), DTD (~7%), SUN397 (~4.6%), and CUB (~3.3%) when compared to CLIP's default prompt. We also design a simple few-shot adapter that learns to choose the best possible sentences to construct generalizable classifiers that outperform the recently proposed CoCoOP by ~2% on average and by over 4% on 4 specialized fine-grained datasets. The code, prompts, and auxiliary text dataset is available at https://github.com/mayug/VDT-Adapter., Comment: Paper accepted at ICCV-W 2023. V2 contains additional comparisons with concurrent works

Event-based cameras are popular for tracking fast-moving objects due to their high temporal resolution, low latency, and high dynamic range. In this paper, we propose a novel algorithm for tracking event blobs using raw events asynchronously in real time. We introduce the concept of an event blob as a spatio-temporal likelihood of event occurrence where the conditional spatial likelihood is blob-like. Many real-world objects such as car headlights or any quickly moving foreground objects generate event blob data. The proposed algorithm uses a nearest neighbour classifier with a dynamic threshold criteria for data association coupled with an extended Kalman filter to track the event blob state. Our algorithm achieves highly accurate blob tracking, velocity estimation, and shape estimation even under challenging lighting conditions and high-speed motions (> 11000 pixels/s). The microsecond time resolution achieved means that the filter output can be used to derive secondary information such as time-to-contact or range estimation, that will enable applications to real-world problems such as collision avoidance in autonomous driving., Comment: 18 pages, 16 figures. The manuscript was accepted on August 7, 2024, by IEEE Transactions on Robotics

Automated shape repair approaches currently lack access to datasets that describe real-world damaged geometry. We present Fantastic Breaks (and Where to Find Them: https://terascale-all-sensing-research-studio.github.io/FantasticBreaks), a dataset containing scanned, waterproofed, and cleaned 3D meshes for 150 broken objects, paired and geometrically aligned with complete counterparts. Fantastic Breaks contains class and material labels, proxy repair parts that join to broken meshes to generate complete meshes, and manually annotated fracture boundaries. Through a detailed analysis of fracture geometry, we reveal differences between Fantastic Breaks and synthetic fracture datasets generated using geometric and physics-based methods. We show experimental shape repair evaluation with Fantastic Breaks using multiple learning-based approaches pre-trained with synthetic datasets and re-trained with subset of Fantastic Breaks., Comment: To be published at CVPR 2023

In this paper, we propose and analyze a new method for online linear quadratic regulator (LQR) control with a priori unknown time-varying cost matrices. The cost matrices are revealed sequentially with the potential for future values to be previewed over a short window. Our novel method involves using the available cost matrices to predict the optimal trajectory, and a tracking controller to drive the system towards it. We adopted the notion of dynamic regret to measure the performance of this proposed online LQR control method, with our main result being that the (dynamic) regret of our method is upper bounded by a constant. Moreover, the regret upper bound decays exponentially with the preview window length, and is extendable to systems with disturbances. We show in simulations that our proposed method offers improved performance compared to other previously proposed online LQR methods., Comment: Submitted to L4DC2023

We investigate the problem of finding paths that enable a robot modeled as a Dubins car (i.e., a constant-speed finite-turn-rate unicycle) to escape from a circular region of space in minimum time. This minimum-time escape problem arises in marine, aerial, and ground robotics in situations where a safety region has been violated and must be exited before a potential negative consequence occurs (e.g., a collision). Using the tools of nonlinear optimal control theory, we show that a surprisingly simple closed-form feedback control law solves this minimum-time escape problem, and that the minimum-time paths have an elegant geometric interpretation., Comment: 7 pages, 5 figures, accepted for 12th IFAC Symposium on Nonlinear Control Systems (NOLCOS)

The degree-restricted random process is a natural algorithmic model for generating graphs with degree sequence D_n=(d_1, \ldots, d_n): starting with an empty n-vertex graph, it sequentially adds new random edges so that the degree of each vertex v_i remains at most d_i. Wormald conjectured in 1999 that, for d-regular degree sequences D_n, the final graph of this process is similar to a uniform random d-regular graph. In this paper we show that, for degree sequences D_n that are not nearly regular, the final graph of the degree-restricted random process differs substantially from a uniform random graph with degree sequence D_n. The combinatorial proof technique is our main conceptual contribution: we adapt the switching method to the degree-restricted process, demonstrating that this enumeration technique can also be used to analyze stochastic processes (rather than just uniform random models, as before)., Comment: 32 pages, 3 figures

Hybridization events complicate the accurate reconstruction of phylogenies, as they lead to patterns of genetic heritability that are unexpected under traditional, bifurcating models of species trees. This has led to the development of methods to infer these varied hybridization events, both methods that reconstruct networks directly, and summary methods that predict individual hybridization events. However, a lack of empirical comparisons between methods - especially pertaining to large networks with varied hybridization scenarios - hinders their practical use. Here, we provide a comprehensive review of popular summary methods: TICR, MSCquartets, HyDe, Patterson's D-Statistic (ABBA-BABA), D3, and Dp. TICR and MSCquartets are based on quartet concordance factors gathered from gene tree topologies and Patterson's D-Statistic, D3, and Dp use site pattern frequencies to identify hybridization events. We then use simulated data to address questions of method accuracy and ideal use scenarios by testing methods against complex networks which depict gene flow events that differ in depth (timing), quantity (single vs. multiple, overlapping hybridizations), and rate of gene flow. We find that deeper or multiple hybridization events may introduce noise and weaken the signal of hybridization, leading to higher false negative rates across methods. Despite some forms of hybridization eluding quartet-based detection methods, MSCquartets displays high precision in most scenarios. While HyDe results in high false negative rates when tested on hybridizations involving ghost lineages, HyDe is the only method to be able to separate hybrid vs parent signals. Lastly, we test the methods on ultraconserved elements from the bee subfamily Nomiinae, finding the possibility of hybridization events between clades which correspond to regions of poor support in the species tree estimated in the original study.

An ensemble of classifiers combines several single classifiers to deliver a final prediction or classification decision. An increasingly provoking question is whether such systems can outperform the single best classifier. If so, what form of an ensemble of classifiers (also known as multiple classifier learning systems or multiple classifiers) yields the most significant benefits in the size or diversity of the ensemble itself? Given that the tests used to detect autism traits are time-consuming and costly, developing a system that will provide the best outcome and measurement of autism spectrum disorder (ASD) has never been critical. In this paper, several single and later multiple classifiers learning systems are evaluated in terms of their ability to predict and identify factors that influence or contribute to ASD for early screening purposes. A dataset of behavioural data and robot-enhanced therapy of 3,000 sessions and 300 hours, recorded from 61 children are utilised for this task. Simulation results show the superior predictive performance of multiple classifier learning systems (especially those with three classifiers per ensemble) compared to individual classifiers, with bagging and boosting achieving excellent results. It also appears that social communication gestures remain the critical contributing factor to the ASD problem among children.

Autonomous Systems (AS) are increasingly proposed, or used, in Safety Critical (SC) applications. Many such systems make use of sophisticated sensor suites and processing to provide scene understanding which informs the AS' decision-making. The sensor processing typically makes use of Machine Learning (ML) and has to work in challenging environments, further the ML-algorithms have known limitations,e.g., the possibility of false-negatives or false-positives in object classification. The well-established safety-analysis methods developed for conventional SC systems are not well-matched to AS, ML, or the sensing systems used by AS. This paper proposes an adaptation of well-established safety-analysis methods to address the specifics of perception-systems for AS, including addressing environmental effects and the potential failure-modes of ML, and provides a rationale for choosing particular sets of guidewords, or prompts, for safety-analysis. It goes on to show how the results of the analysis can be used to inform the design and verification of the AS and illustrates the new method by presenting a partial analysis of a road vehicle. Illustrations in the paper are primarily based on optical sensing, however the paper discusses the applicability of the method to other sensing modalities and its role in a wider safety process addressing the overall capabilities of AS., Comment: 53 pages, 9 figures, 19 tables

In this work, photometric and spectroscopic analyses of a very low-luminosity Type IIb supernova (SN) 2016iyc have been performed. SN 2016iyc lies near the faint end among the distribution of similar supernovae (SNe). Given lower ejecta mass ($M_{\rm ej}$) and low nickel mass ($M_{\rm Ni}$) from the literature, combined with SN 2016iyc lying near the faint end, one-dimensional stellar evolution models of 9 - 14 M$_{\odot}$ zero-age main-sequence (ZAMS) stars as the possible progenitors of SN 2016iyc have been performed using the publicly available code MESA. Moreover, synthetic explosions of the progenitor models have been simulated using the hydrodynamic evolution codes STELLA and SNEC. The bolometric luminosity light curve and photospheric velocities produced through synthetic explosions of ZAMS stars of mass in the range 12 - 13 M$_{\odot}$ having a pre-supernova radius $R_{\mathrm{0}} =$ (240 - 300) R$_{\odot}$, with $M_{\rm ej} =$ (1.89 - 1.93) M$_{\odot}$, explosion energy $E_{\rm exp} = $ (0.28 - 0.35) $\times 10^{51}$ erg, and $M_{\rm Ni} < 0.09$ M$_{\odot}$, are in good agreement with observations; thus, SN 2016iyc probably exploded from a progenitor near the lower mass limits for SNe IIb. Finally, hydrodynamic simulations of the explosions of SN 2016gkg and SN 2011fu have also been performed to compare intermediate- and high-luminosity examples among well-studied SNe IIb. The results of progenitor modelling and synthetic explosions for SN 2016iyc, SN 2016gkg, and SN 2011fu exhibit a diverse range of mass for the possible progenitors of SNe IIb., Comment: Final updated version. Accepted for publication in MNRAS

We present BVRI and unfiltered Clear light curves of 70 stripped-envelope supernovae (SESNe), observed between 2003 and 2020, from the Lick Observatory Supernova Search (LOSS) follow-up program. Our SESN sample consists of 19 spectroscopically normal SNe~Ib, two peculiar SNe Ib, six SN Ibn, 14 normal SNe Ic, one peculiar SN Ic, ten SNe Ic-BL, 15 SNe IIb, one ambiguous SN IIb/Ib/c, and two superluminous SNe. Our follow-up photometry has (on a per-SN basis) a mean coverage of 81 photometric points (median of 58 points) and a mean cadence of 3.6d (median of 1.2d). From our full sample, a subset of 38 SNe have pre-maximum coverage in at least one passband, allowing for the peak brightness of each SN in this subset to be quantitatively determined. We describe our data collection and processing techniques, with emphasis toward our automated photometry pipeline, from which we derive publicly available data products to enable and encourage further study by the community. Using these data products, we derive host-galaxy extinction values through the empirical colour evolution relationship and, for the first time, produce accurate rise-time measurements for a large sample of SESNe in both optical and infrared passbands. By modeling multiband light curves, we find that SNe Ic tend to have lower ejecta masses and lower ejecta velocities than SNe~Ib and IIb, but higher $^{56}$Ni masses., Comment: Accepted by MNRAS

Programming is a powerful and ubiquitous problem-solving tool. Developing systems that can assist programmers or even generate programs independently could make programming more productive and accessible, yet so far incorporating innovations in AI has proven challenging. Recent large-scale language models have demonstrated an impressive ability to generate code, and are now able to complete simple programming tasks. However, these models still perform poorly when evaluated on more complex, unseen problems that require problem-solving skills beyond simply translating instructions into code. For example, competitive programming problems which require an understanding of algorithms and complex natural language remain extremely challenging. To address this gap, we introduce AlphaCode, a system for code generation that can create novel solutions to these problems that require deeper reasoning. In simulated evaluations on recent programming competitions on the Codeforces platform, AlphaCode achieved on average a ranking of top 54.3% in competitions with more than 5,000 participants. We found that three key components were critical to achieve good and reliable performance: (1) an extensive and clean competitive programming dataset for training and evaluation, (2) large and efficient-to-sample transformer-based architectures, and (3) large-scale model sampling to explore the search space, followed by filtering based on program behavior to a small set of submissions., Comment: 74 pages

We investigate partially observed Markov decision processes (POMDPs) with cost functions regularized by entropy terms describing state, observation, and control uncertainty. Standard POMDP techniques are shown to offer bounded-error solutions to these entropy-regularized POMDPs, with exact solutions possible when the regularization involves the joint entropy of the state, observation, and control trajectories. Our joint-entropy result is particularly surprising since it constitutes a novel, tractable formulation of active state estimation., Comment: 20 pages, 2 figures, submitted

The rapid rise in demand for training large neural network architectures has brought into focus the need for partitioning strategies, for example by using data, model, or pipeline parallelism. Implementing these methods is increasingly supported through program primitives, but identifying efficient partitioning strategies requires expensive experimentation and expertise. We present the prototype of an automated partitioner that seamlessly integrates into existing compilers and existing user workflows. Our partitioner enables SPMD-style parallelism that encompasses data parallelism and parameter/activation sharding. Through a combination of inductive tactics and search in a platform-independent partitioning IR, automap can recover expert partitioning strategies such as Megatron sharding for transformer layers., Comment: Workshop on ML for Systems at NeurIPS 2021

A growing body of research documents the effectiveness of classroom management programs on a range of student outcomes, yet few early-career teachers receive training on these practices prior to entering the classroom. Moreover, few studies have attended to how variations in teacher distress or level of classroom misbehavior affects training benefits. This study reports findings from a randomized trial of a teacher training program that combined two evidence-based programs (Good Behavior Game [GBG] and MyTeachingPartnerTM [MTP]) to determine their impact on novice teachers and their students. In addition, the current study reports findings on moderated impacts by initial teacher distress as well as the overall classroom level of misbehavior. The sample included 188 early-career teachers (grades K-3) in their first three years of teaching from three large, urban school districts. Analyses indicated that intervention had no main effects, but yielded moderated impact depending on the combination of the baseline levels of classroom disruptive behavior and teacher distress; it appears that the program impacts were greatest in the highest risk circumstance (i.e., high teacher stress and elevated challenging student behaviors). For those classrooms, those assigned to intervention evidence improved behavior and student achievement compared to control counterparts by the spring of the training year, relative to the fall baseline (d = 0.18-0.70 depending on outcome). This study is significant in that it highlights effects during a critical window of training and coaching for early career teachers and the need to consider teacher and classroom contextual factors that may moderate professional development efforts. [This paper was published in "Journal of School Psychology" v78 p75-95 2020.]

We study the problem of controlling a partially observed Markov decision process (POMDP) to either aid or hinder the estimation of its state trajectory. We encode the estimation objectives via the smoother entropy, which is the conditional entropy of the state trajectory given measurements and controls. Consideration of the smoother entropy contrasts with previous approaches that instead resort to marginal (or instantaneous) state entropies due to tractability concerns. By establishing novel expressions for the smoother entropy in terms of the POMDP belief state, we show that both the problems of minimising and maximising the smoother entropy in POMDPs can surprisingly be reformulated as belief-state Markov decision processes with concave cost and value functions. The significance of these reformulations is that they render the smoother entropy a tractable optimisation objective, with structural properties amenable to the use of standard POMDP solution techniques for both active estimation and obfuscation. Simulations illustrate that optimisation of the smoother entropy leads to superior trajectory estimation and obfuscation compared to alternative approaches., Comment: 41 pages, 3 figures, accepted for publication in IEEE Transactions on Automatic Control

The adaptable choosability of a multigraph $G$, denoted $\mathrm{ch}_a(G)$, is the smallest integer $k$ such that any edge labelling, $\tau$, of $G$ and any assignment of lists of size $k$ to the vertices of $G$ permits a list colouring, $\sigma$, of $G$ such that there is no edge $e = uv$ where $\tau(e) = \sigma(u) = \sigma(v)$. Here we show that for a multigraph $G$ with maximum degree $\Delta$ and no cycles of length 3 or 4, $\mathrm{ch}_a(G) \leq (2\sqrt{2}+o(1))\sqrt{\Delta/\ln\Delta}$. Under natural restrictions we can show that the same bound holds for the conflict choosability of $G$, which is a closely related parameter defined by Dvo\v{r}\'ak, Esperet, Kang and Ozeki [arXiv:1803.10962]., Comment: 30 pages

In this paper, we consider the problem of controlling a partially observed Markov decision process (POMDP) in order to actively estimate its state trajectory over a fixed horizon with minimal uncertainty. We pose a novel active smoothing problem in which the objective is to directly minimise the smoother entropy, that is, the conditional entropy of the (joint) state trajectory distribution of concern in fixed-interval Bayesian smoothing. Our formulation contrasts with prior active approaches that minimise the sum of conditional entropies of the (marginal) state estimates provided by Bayesian filters. By establishing a novel form of the smoother entropy in terms of the POMDP belief (or information) state, we show that our active smoothing problem can be reformulated as a (fully observed) Markov decision process with a value function that is concave in the belief state. The concavity of the value function is of particular importance since it enables the approximate solution of our active smoothing problem using piecewise-linear function approximations in conjunction with standard POMDP solvers. We illustrate the approximate solution of our active smoothing problem in simulation and compare its performance to alternative approaches based on minimising marginal state estimate uncertainties., Comment: 7 pages, 3 figures, accepted for presentation at 2021 European Control Conference

In this paper we investigate the problem of controlling a partially observed stochastic dynamical system such that its state is difficult to infer using a (fixed-interval) Bayesian smoother. This problem arises naturally in applications in which it is desirable to keep the entire state trajectory of a system concealed. We pose our smoothing-averse control problem as the problem of maximising the (joint) entropy of smoother state estimates (i.e., the joint conditional entropy of the state trajectory given the history of measurements and controls). We show that the entropy of Bayesian smoother estimates for general nonlinear state-space models can be expressed as the sum of entropies of marginal state estimates given by Bayesian filters. This novel additive form allows us to reformulate the smoothing-averse control problem as a fully observed stochastic optimal control problem in terms of the usual concept of the information (or belief) state, and solve the resulting problem via dynamic programming. We illustrate the applicability of smoothing-averse control to privacy in cloud-based control and covert robotic navigation., Comment: 8 pages, 2 figures, accepted for presentation at 2021 American Control Conference

Verifiable training has shown success in creating neural networks that are provably robust to a given amount of noise. However, despite only enforcing a single robustness criterion, its performance scales poorly with dataset complexity. On CIFAR10, a non-robust LeNet model has a 21.63% error rate, while a model created using verifiable training and a L-infinity robustness criterion of 8/255, has an error rate of 57.10%. Upon examination, we find that when labeling visually similar classes, the model's error rate is as high as 61.65%. We attribute the loss in performance to inter-class similarity. Similar classes (i.e., close in the feature space) increase the difficulty of learning a robust model. While it's desirable to train a robust model for a large robustness region, pairwise class similarities limit the potential gains. Also, consideration must be made regarding the relative cost of mistaking similar classes. In security or safety critical tasks, similar classes are likely to belong to the same group, and thus are equally sensitive. In this work, we propose a new approach that utilizes inter-class similarity to improve the performance of verifiable training and create robust models with respect to multiple adversarial criteria. First, we use agglomerate clustering to group similar classes and assign robustness criteria based on the similarity between clusters. Next, we propose two methods to apply our approach: (1) Inter-Group Robustness Prioritization, which uses a custom loss term to create a single model with multiple robustness guarantees and (2) neural decision trees, which trains multiple sub-classifiers with different robustness guarantees and combines them in a decision tree architecture. On Fashion-MNIST and CIFAR10, our approach improves clean performance by 9.63% and 30.89% respectively. On CIFAR100, our approach improves clean performance by 26.32%., Comment: Acceped at AAAI21

A key challenge in visual place recognition (VPR) is recognizing places despite drastic visual appearance changes due to factors such as time of day, season, weather or lighting conditions. Numerous approaches based on deep-learnt image descriptors, sequence matching, domain translation, and probabilistic localization have had success in addressing this challenge, but most rely on the availability of carefully curated representative reference images of the possible places. In this paper, we propose a novel approach, dubbed Bayesian Selective Fusion, for actively selecting and fusing informative reference images to determine the best place match for a given query image. The selective element of our approach avoids the counterproductive fusion of every reference image and enables the dynamic selection of informative reference images in environments with changing visual conditions (such as indoors with flickering lights, outdoors during sunshowers or over the day-night cycle). The probabilistic element of our approach provides a means of fusing multiple reference images that accounts for their varying uncertainty via a novel training-free likelihood function for VPR. On difficult query images from two benchmark datasets, we demonstrate that our approach matches and exceeds the performance of several alternative fusion approaches along with state-of-the-art techniques that are provided with prior (unfair) knowledge of the best reference images. Our approach is well suited for long-term robot autonomy where dynamic visual environments are commonplace since it is training-free, descriptor-agnostic, and complements existing techniques such as sequence matching., Comment: 8 pages, 10 figures, accepted in the IEEE Robotics and Automation Letters

This paper considers the quickest detection problem for hidden Markov models (HMMs) in a Bayesian setting. We construct an augmented HMM representation of the problem that allows the application of a dynamic programming approach to prove that Shiryaev's rule is an (exact) optimal solution. This augmented representation highlights the problem's fundamental information structure and suggests possible relaxations to more exotic change event priors not appearing in the literature. Finally, this augmented representation also allows us to present an efficient computational method for implementing the optimal solution.

Adversarial examples are a type of attack on machine learning (ML) systems which cause misclassification of inputs. Achieving robustness against adversarial examples is crucial to apply ML in the real world. While most prior work on adversarial examples is empirical, a recent line of work establishes fundamental limitations of robust classification based on cryptographic hardness. Most positive and negative results in this field however assume that there is a fixed target metric which constrains the adversary, and we argue that this is often an unrealistic assumption. In this work we study the limitations of robust classification if the target metric is uncertain. Concretely, we construct a classification problem, which admits robust classification by a small classifier if the target metric is known at the time the model is trained, but for which robust classification is impossible for small classifiers if the target metric is chosen after the fact. In the process, we explore a novel connection between hardness of robust classification and bounded storage model cryptography., Comment: 25 pages, 1 figure, under submission, fixe typos from previous version

Backdoor attacks mislead machine-learning models to output an attacker-specified class when presented a specific trigger at test time. These attacks require poisoning the training data to compromise the learning algorithm, e.g., by injecting poisoning samples containing the trigger into the training set, along with the desired class label. Despite the increasing number of studies on backdoor attacks and defenses, the underlying factors affecting the success of backdoor attacks, along with their impact on the learning algorithm, are not yet well understood. In this work, we aim to shed light on this issue by unveiling that backdoor attacks induce a smoother decision function around the triggered samples -- a phenomenon which we refer to as \textit{backdoor smoothing}. To quantify backdoor smoothing, we define a measure that evaluates the uncertainty associated to the predictions of a classifier around the input samples. Our experiments show that smoothness increases when the trigger is added to the input samples, and that this phenomenon is more pronounced for more successful attacks. We also provide preliminary evidence that backdoor triggers are not the only smoothing-inducing patterns, but that also other artificial patterns can be detected by our approach, paving the way towards understanding the limitations of current defenses and designing novel ones., Comment: 9 pages, 7 figures, under submission

In this paper, we consider the problem of computing parameters of an objective function for a discrete-time optimal control problem from state and control trajectories with active control constraints. We propose a novel method of inverse optimal control that has a computationally efficient online form in which pairs of states and controls from given state and control trajectories are processed sequentially without being stored or processed in batches. We establish conditions guaranteeing the uniqueness of the objective-function parameters computed by our proposed method from trajectories with active control constraints. We illustrate our proposed method in simulation., Comment: 10 pages, 4 figures, Accepted for publication in Automatica

The problem of quickly diagnosing an unknown change in a stochastic process is studied. We establish novel bounds on the performance of misspecified diagnosis algorithms designed for changes that differ from those of the process, and pose and solve a new robust quickest change diagnosis problem in the asymptotic regime of few false alarms and false isolations. Simulations suggest that our asymptotically robust solution offers a computationally efficient alternative to generalised likelihood ratio algorithms., Comment: 19 pages, 2 figures, Accepted for publication in IEEE Transactions on Automatic Control

This paper presents a novel collision avoidance strategy for unmanned aircraft detect and avoid that requires only information about the relative bearing angle between an aircraft and hazard. It is shown that this bearing-only strategy can be conceived as the solution to a novel differential game formulation of collision avoidance, and has several intuitive properties including maximising the instantaneous range acceleration in situations where the hazard is stationary or has a finite turn rate. The performance of the bearing-only strategy is illustrated in simulations based on test cases drawn from draft minimum operating performance standards for unmanned aircraft detect and avoid systems., Comment: 35 pages, 12 figures, Accepted for publication in the Journal of Guidance, Control and Dynamics

Standardisation is an important component in the maturation of any field of technology. It contributes to the formation of a recognisable identity and enables interactions with a wider community. This article reviews past and current standardisation initiatives in the field of Open Source Hardware (OSH). While early initiatives focused on aspects such as licencing, intellectual property and documentation formats, recent efforts extend to ways for users to exercise their rights under open licences and to keep OSH projects discoverable and accessible online. We specifically introduce two standards that are currently being released and call for early users and contributors, the DIN SPEC 3105 and the Open Know How Manifest Specification. Finally, we reflect on challenges around standardisation in the community and relevant areas for future development such as an open tool chain, modularity and hardware specific interface standards., Comment: 9 Pages without abstract and references (else 13), no figures

Within national and international innovation systems a pandemic calls for large-scale action by many actors across sectors, to mobilise resources, developing and manufacturing Crisis-Critical Products (CC-Products) efficiently and in the huge quantities needed. Nowadays, this also includes digital innovations from complex epidemiological models, AI, to open data platforms for prevention, diagnostic and treatment. Amongst the many challenges during a pandemic, innovation and manufacturing stakeholders find themselves engaged in new relationships, and are likely to face intellectual property (IP) related challenges. This paper adopts an IP perspective on the COVID-19 pandemic to identify pandemic related IP considerations and IP challenges. The focus is on challenges related to research, development and urgent upscaling of capacity to manufacture CC-Products in the huge volumes suddenly in demand. Its purpose is to provide a structure for steering clear of IP challenges to avoid delays in fighting a pandemic. We identify 4 stakeholder groups concerned with IP challenges: (i) governments, (ii) organisations owning existing Crisis-Critical IP, described as incumbents in Crisis-Critical Sectors (CC-Sectors), (iii) manufacturing firms from other sectors normally not producing CC-Products suddenly rushing into CC-Sectors to support the manufacturing of CC-Products (new entrants), and (iv) voluntary grassroot initiatives that are formed during a pandemic. This paper discusses IP challenges related to the development and manufacturing of technologies and products for (i) prevention (of spread), (ii) diagnosis of infected patients and (iii) the development of treatments. We offer an initial discussion of potential response measures to reduce IP associated risks among industrial stakeholders during a pandemic., Comment: 18 pages, 4 figures, 3 tables

In this paper, beam diagnostic and monitoring tools developed by the MAX IV Operations Group are discussed. In particular, new beam position monitoring and accelerator tunes visualization software tools, as well as tools that directly influence the beam quality and stability are introduced. An availability and downtime monitoring application is also presented.

In this work, BV RI light curves of 55 Type II supernovae (SNe II) from the Lick Observatory Supernova Search program obtained with the Katzman Automatic Imaging Telescope and the 1 m Nickel telescope from 2006 to 2018 are presented. Additionally, more than 150 spectra gathered with the 3 m Shane telescope are published. We conduct an analyse of the peak absolute magnitudes, decline rates, and time durations of different phases of the light and colour curves. Typically, our light curves are sampled with a median cadence of 5.5 days for a total of 5093 photometric points. In average V-band plateau declines with a rate of 1.29 mag (100 days)-1, which is consistent with previously published samples. For each band, the plateau slope correlates with the plateau length and the absolute peak magnitude: SNe II with steeper decline have shorter plateau duration and are brighter. A time-evolution analysis of spectral lines in term of velocities and pseudoequivalent widths is also presented in this paper. Our spectroscopic sample ranges between 1 and 200 days post-explosion and has a median ejecta expansion velocity at 50 days post-explosion of 6500 km/s (Halpha line) and a standard dispersion of 2000 km/s. Nebular spectra are in good agreement with theoretical models using a progenitor star having a mass <16 Msol. All the data are available to the community and will help to understand SN II diversity better, and therefore to improve their utility as cosmological distance indicators., Comment: 24 pages, 12 figures, accepted for publication in MNRAS

We present BVRI and unfiltered light curves of 93 Type Ia supernovae (SNe Ia) from the Lick Observatory Supernova Search (LOSS) follow-up program conducted between 2005 and 2018. Our sample consists of 78 spectroscopically normal SNe Ia, with the remainder divided between distinct subclasses (three SN 1991bg-like, three SN 1991T-like, four SNe Iax, two peculiar, and three super-Chandrasekhar events), and has a median redshift of 0.0192. The SNe in our sample have a median coverage of 16 photometric epochs at a cadence of 5.4 days, and the median first observed epoch is ~4.6 days before maximum B-band light. We describe how the SNe in our sample are discovered, observed, and processed, and we compare the results from our newly developed automated photometry pipeline to those from the previous processing pipeline used by LOSS. After investigating potential biases, we derive a final systematic uncertainty of 0.03 mag in BVRI for our dataset. We perform an analysis of our light curves with particular focus on using template fitting to measure the parameters that are useful in standardising SNe Ia as distance indicators. All of the data are available to the community, and we encourage future studies to incorporate our light curves in their analyses., Comment: 29 pages, 13 figures, accepted for publication in MNRAS

The cochromatic number $Z(G)$ of a graph $G$ is the fewest number of colors needed to color the vertices of $G$ so that each color class is a clique or an independent set. In a fractional cocoloring of $G$ a non-negative weight is assigned to each clique and independent set so that for each vertex $v$, the sum of the weights of all cliques and independent sets containing $v$ is at least one. The smallest total weight of such a fractional cocoloring of $G$ is the fractional cochromatic number $Z_f(G)$. In this paper we prove results for the fractional cochromatic number $Z_f(G)$ that parallel results for $Z(G)$ and the well studied fractional chromatic number $\chi_f{(G)}$. For example $Z_f(G)=\chi_f(G)$ when $G$ is triangle-free, except when the only nontrivial component of $G$ is a star. More generally, if $G$ contains no $k$-clique, then $Z_f(G)\le \chi_f(G)\le Z_f(G)+R(k,k)$. Moreover, every graph $G$ with $\chi_f(G)=m$ contains a subgraph $H$ with $Z_f(H)\ge (\frac 14 - o(1))\frac m{\log_2 m}$. We also prove that the maximum value of $Z_f(G)$ over all graphs $G$ of order $n$ is $\Theta (n/\log n)$, and the maximum over all graphs embedded on an orientable surface of genus $g$ is $\Theta(\sqrt g / \log g)$.

Motivation: Due to the central role of protein structure in molecular recognition, great computational efforts are devoted to modeling protein structures and motions that mediate structural rearrangements. The size, dimensionality, and non-linearity of the protein structure space present outstanding challenges. Such challenges also arise in robot motion planning, and robotics-inspired treatments of protein structure and motion are increasingly showing high exploration capability. Encouraged by such findings, we debut here ROMEO, which stands for Robotics prOtein Motion ExplOration framework. ROMEO is an open-source, object-oriented platform that allows researchers access to and reproducibility of published robotics-inspired algorithms for modeling protein structures and motions, as well as facilitates novel algorithmic design via its plug-and-play architecture. Availability and implementation: ROMEO is written in C++ and is available in GitLab (https://github.com/). This software is freely available under the Creative Commons license (Attribution and Non-Commercial). Contact: amarda@gmu.edu, Comment: 6 pages, 5 figures