Search

Chaining functions for longer workloads is a key use case for FaaS platforms in data applications. However, modern data pipelines differ significantly from typical serverless use cases (e.g., webhooks and microservices); this makes it difficult to retrofit existing pipeline frameworks due to structural constraints. In this paper, we describe these limitations in detail and introduce bauplan, a novel FaaS programming model and serverless runtime designed for data practitioners. bauplan enables users to declaratively define functional Directed Acyclic Graphs (DAGs) along with their runtime environments, which are then efficiently executed on cloud-based workers. We show that bauplan achieves both better performance and a superior developer experience for data workloads by making the trade-off of reducing generality in favor of data-awareness, Comment: Accepted for the 10th International Workshop on Serverless Computing (pre-print)

Artificial engineering often relies on scarce materials, possessing ecological, social, and political risks. In contrast, nature achieves advanced functionalities using only abundant elements and water. Despite this simple insight, incorporating nature-inspired devices made from readily available resources into industrial applications remains a formidable challenge. Here, we present a sustainable 'blue battery'-a water-based energy storage device that mimics natural processes and reduces dependence on scarce materials. This battery harnesses the distinctive electrical properties of water confined within nanometer-scale pores, enabling construction from abundant, non-toxic materials using scalable nanofabrication techniques. It operates without unwanted side reactions, maintains nearly 100% efficiency after 60,000 charge-discharge cycles, and supports a voltage window of up to 1.65 V while achieving competitive power and energy density. The core component, a van der Waals clay nanostructure, uses water as the sole electrolyte with enhanced activity due to interactions with its 2D surface. This water-driven technology offers a sustainable platform for energy systems, with potential applications from small-scale electronics to global energy infrastructure-even in extreme environments like Mars. Our work advances nature-inspired energy solutions, opening new possibilities for the sustainable engineering of blue devices., Comment: Main text: 17 pages, 4 figures, 50 references Suplementary: 29 pages, 26 figures, 3 tables, 3 references

Solar redox flow batteries (SRFB) have received much attention as an alternative integrated technology for simultaneous conversion and storage of solar energy. Yet, the photocatalytic efficiency of semiconductor-based single photoelectrode, such as hematite, remains low due to the trade-off between fast electron hole recombination and insufficient light utilization, as well as inferior reaction kinetics at the solid/liquid interface. Herein, we present an {\alpha}-Fe2O3/CuxO p-n junction, coupled with a readily scalable nanostructure, that increases the electrochemically active sites and improves charge separation. Thanks to light-assisted scanning electrochemical microscopy (Photo-SECM), we elucidate the morphology-dependent carrier transfer process involved in the photo-oxidation reaction at a {\alpha}-Fe2O3 photoanode. The optimized nanostructured is then exploited in the {\alpha}-Fe2O3/CuxO p-n junction, achieving an outstanding unbiased photocurrent density of 0.46 mA/cm2, solar-to-chemical (STC) efficiency over 0.35% and a stable photocharge-discharge cycling. The average solar-to-output energy efficiency (SOEE) for this unassisted {\alpha}-Fe2O3-based SRFB system reaches 0.18%, comparable to previously reported DSSC-assisted hematite SRFBs. The use of earth-abundant materials and the compatibility with scalable nanostructuring and heterojunction preparation techniques, offer promising opportunities for cost-effective device deployment in real-world applications.

Circular Dichroism (CD) can distinguish the handedness of chiral molecules. However, it is typically very weak due to vanishing absorption at low molecular concentrations. Here, we suggest Thermal Circular Dichroism (TCD) for chiral detection, leveraging the temperature difference in the chiral sample when subjected to right and left-circularly polarized excitations. The TCD combines the enantiospecificity of circular dichroism with the higher sensitivity of thermal measurements, while introducing new opportunities in the thermal domain that can be synergistically combined with optical approaches. We propose a theoretical framework to understand the TCD of individual and arrays of resonators covered by chiral molecules. To enhance the weak TCD of chiral samples, we first use individual dielectric Mie resonators and identify chirality transfer and self-heating as the underlying mechanisms giving rise to the differential temperature. However, inherent limitations imposed by the materials and geometries of such resonators make it challenging to surpass a certain level in enhancements. To overcome this, we suggest nonlocal thermal and electromagnetic interactions in arrays. We predict that a combination of chirality transfer to Mie resonators, collective thermal effects, and optical lattice resonance could, in principle, offer more than 4 orders of magnitude enhancement in TCD. Our thermonanophotonic-based approach thus establishes key concepts for ultrasensitive chiral detection.

Exploring nonequilibrium hot carriers from plasmonic metal nanostructures is a dynamic field in optoelectronics, driving photochemical reactions such as solar fuel generation. The hot carrier injection mechanism and the reaction rate are highly impacted by the metal/molecule interaction. However, determining the primary type of the reaction and thus the injection mechanism of the hot carriers has remained elusive. In this work, we reveal an electron injection mechanism deviating from a purely outersphere process for the reduction of ferricyanide redox molecule in a gold/p-type gallium nitride (Au/p- GaN) photocathode system. Combining our experimental approach with ab initio simulations, we discover that the efficient inner-sphere transfer of low-energy electrons leads to a continuous enhancement in the photocathode device performance in the interband regime. These findings provide important mechanistic insights, showing our methodology as a powerful tool for analyzing and engineering hot-carrier-driven processes in plasmonic photocatalytic systems and optoelectronic devices.

In decentralized multiagent trajectory planners, agents need to communicate and exchange their positions to generate collision-free trajectories. However, due to localization errors/uncertainties, trajectory deconfliction can fail even if trajectories are perfectly shared between agents. To address this issue, we first present PARM and PARM*, perception-aware, decentralized, asynchronous multiagent trajectory planners that enable a team of agents to navigate uncertain environments while deconflicting trajectories and avoiding obstacles using perception information. PARM* differs from PARM as it is less conservative, using more computation to find closer-to-optimal solutions. While these methods achieve state-of-the-art performance, they suffer from high computational costs as they need to solve large optimization problems onboard, making it difficult for agents to replan at high rates. To overcome this challenge, we present our second key contribution, PRIMER, a learning-based planner trained with imitation learning (IL) using PARM* as the expert demonstrator. PRIMER leverages the low computational requirements at deployment of neural networks and achieves a computation speed up to 5500 times faster than optimization-based approaches., Comment: 7 pages, 3 figures

Automation holds the potential to assist surgeons in robotic interventions, shifting their mental work load from visuomotor control to high level decision making. Reinforcement learning has shown promising results in learning complex visuomotor policies, especially in simulation environments where many samples can be collected at low cost. A core challenge is learning policies in simulation that can be deployed in the real world, thereby overcoming the sim-to-real gap. In this work, we bridge the visual sim-to-real gap with an image-based reinforcement learning pipeline based on pixel-level domain adaptation and demonstrate its effectiveness on an image-based task in deformable object manipulation. We choose a tissue retraction task because of its importance in clinical reality of precise cancer surgery. After training in simulation on domain-translated images, our policy requires no retraining to perform tissue retraction with a 50% success rate on the real robotic system using raw RGB images. Furthermore, our sim-to-real transfer method makes no assumptions on the task itself and requires no paired images. This work introduces the first successful application of visual sim-to-real transfer for robotic manipulation of deformable objects in the surgical field, which represents a notable step towards the clinical translation of cognitive surgical robotics.

Traditional optimization-based planners, while effective, suffer from high computational costs, resulting in slow trajectory generation. A successful strategy to reduce computation time involves using Imitation Learning (IL) to develop fast neural network (NN) policies from those planners, which are treated as expert demonstrators. Although the resulting NN policies are effective at quickly generating trajectories similar to those from the expert, (1) their output does not explicitly account for dynamic feasibility, and (2) the policies do not accommodate changes in the constraints different from those used during training. To overcome these limitations, we propose Constraint-Guided Diffusion (CGD), a novel IL-based approach to trajectory planning. CGD leverages a hybrid learning/online optimization scheme that combines diffusion policies with a surrogate efficient optimization problem, enabling the generation of collision-free, dynamically feasible trajectories. The key ideas of CGD include dividing the original challenging optimization problem solved by the expert into two more manageable sub-problems: (a) efficiently finding collision-free paths, and (b) determining a dynamically-feasible time-parametrization for those paths to obtain a trajectory. Compared to conventional neural network architectures, we demonstrate through numerical evaluations significant improvements in performance and dynamic feasibility under scenarios with new constraints never encountered during training., Comment: 8 pages, 3 figures

As the Lakehouse architecture becomes more widespread, ensuring the reproducibility of data workloads over data lakes emerges as a crucial concern for data engineers. However, achieving reproducibility remains challenging. The size of data pipelines contributes to slow testing and iterations, while the intertwining of business logic and data management complicates debugging and increases error susceptibility. In this paper, we highlight recent advancements made at Bauplan in addressing this challenge. We introduce a system designed to decouple compute from data management, by leveraging a cloud runtime alongside Nessie, an open-source catalog with Git semantics. Demonstrating the system's capabilities, we showcase its ability to offer time-travel and branching semantics on top of object storage, and offer full pipeline reproducibility with a few CLI commands., Comment: Pre-print of paper accepted at SIGMOD (DEEM2024)

Scattering-type scanning near-field optical microscope (s-SNOM) has become an essential tool to study polaritons - quasiparticles of light coupled to collective charge oscillations - via direct probing of their near field with a spatial resolution far beyond the diffraction limit. However, extraction of the polariton complex propagation constant from the near-field images requires subtle considerations that have not received necessary attention so far. In this study, we discuss important yet overlooked aspects of the near-field analysis. First, we experimentally demonstrate that the sample orientation inside the s-SNOM may significantly affect the near-field interference pattern of mid-infrared polaritons, leading to an error in momentum measurement up to 7.7% even for the modes with effective index of 12.5. Second, we establish a methodology to correctly extract the polariton damping rate from the interference fringes depending on their origin - the s-SNOM nano-tip or the material edge. Overall, our work provides a unified framework for the accurate extraction of the polariton momentum and damping from the near-field interference fringes., Comment: 15 pages, 4 figures

Negotiation is the basis of social interactions; humans negotiate everything from the price of cars to how to share common resources. With rapidly growing interest in using large language models (LLMs) to act as agents on behalf of human users, such LLM agents would also need to be able to negotiate. In this paper, we study how well LLMs can negotiate with each other. We develop NegotiationArena: a flexible framework for evaluating and probing the negotiation abilities of LLM agents. We implemented three types of scenarios in NegotiationArena to assess LLM's behaviors in allocating shared resources (ultimatum games), aggregate resources (trading games) and buy/sell goods (price negotiations). Each scenario allows for multiple turns of flexible dialogues between LLM agents to allow for more complex negotiations. Interestingly, LLM agents can significantly boost their negotiation outcomes by employing certain behavioral tactics. For example, by pretending to be desolate and desperate, LLMs can improve their payoffs by 20\% when negotiating against the standard GPT-4. We also quantify irrational negotiation behaviors exhibited by the LLM agents, many of which also appear in humans. Together, \NegotiationArena offers a new environment to investigate LLM interactions, enabling new insights into LLM's theory of mind, irrationality, and reasoning abilities.

Imitation learning (IL) can train computationally-efficient sensorimotor policies from a resource-intensive Model Predictive Controller (MPC), but it often requires many samples, leading to long training times or limited robustness. To address these issues, we combine IL with a variant of robust MPC that accounts for process and sensing uncertainties, and we design a data augmentation (DA) strategy that enables efficient learning of vision-based policies. The proposed DA method, named Tube-NeRF, leverages Neural Radiance Fields (NeRFs) to generate novel synthetic images, and uses properties of the robust MPC (the tube) to select relevant views and to efficiently compute the corresponding actions. We tailor our approach to the task of localization and trajectory tracking on a multirotor, by learning a visuomotor policy that generates control actions using images from the onboard camera as only source of horizontal position. Numerical evaluations show 80-fold increase in demonstration efficiency and a 50% reduction in training time over current IL methods. Additionally, our policies successfully transfer to a real multirotor, achieving low tracking errors despite large disturbances, with an onboard inference time of only 1.5 ms. Video: https://youtu.be/_W5z33ZK1m4, Comment: Video: https://youtu.be/_W5z33ZK1m4. Evolved paper from our previous work: arXiv:2210.10127

Applications in photodetection, photochemistry, and active metamaterials and metasurfaces require fundamental understanding of ultrafast nonthermal and thermal electron processes in metallic nanosystems. Significant progress has been recently achieved in synthesis and investigation of low-loss monocrystalline gold, opening up opportunities for its use in ultrathin nanophotonic architectures. Here, we reveal fundamental differences in hot-electron thermalisation dynamics between monocrystalline and polycrystalline ultrathin (down to 10 nm thickness) gold films. Comparison of weak and strong excitation regimes showcases a counterintuitive unique interplay between thermalised and non-thermalised electron dynamics in mesoscopic gold with the important influence of the X-point interband transitions on the intraband electron relaxation. We also experimentally demonstrate the effect of hot-electron transfer into a substrate and the substrate thermal properties on electron-electron and electron-phonon scattering in ultrathin films. The hot-electron injection efficiency from monocrystalline gold into TiO2, approaching 9% is measured, close to the theoretical limit. These experimental and modelling results reveal the important role of crystallinity and interfaces on the microscopic electronic processes important in numerous applications., Comment: 9 pages, 4 figures

Fully decentralized, multiagent trajectory planners enable complex tasks like search and rescue or package delivery by ensuring safe navigation in unknown environments. However, deconflicting trajectories with other agents and ensuring collision-free paths in a fully decentralized setting is complicated by dynamic elements and localization uncertainty. To this end, this paper presents (1) an uncertainty-aware multiagent trajectory planner and (2) an image segmentation-based frame alignment pipeline. The uncertainty-aware planner propagates uncertainty associated with the future motion of detected obstacles, and by incorporating this propagated uncertainty into optimization constraints, the planner effectively navigates around obstacles. Unlike conventional methods that emphasize explicit obstacle tracking, our approach integrates implicit tracking. Sharing trajectories between agents can cause potential collisions due to frame misalignment. Addressing this, we introduce a novel frame alignment pipeline that rectifies inter-agent frame misalignment. This method leverages a zero-shot image segmentation model for detecting objects in the environment and a data association framework based on geometric consistency for map alignment. Our approach accurately aligns frames with only 0.18 m and 2.7 deg of mean frame alignment error in our most challenging simulation scenario. In addition, we conducted hardware experiments and successfully achieved 0.29 m and 2.59 deg of frame alignment error. Together with the alignment framework, our planner ensures safe navigation in unknown environments and collision avoidance in decentralized settings., Comment: 7 pages, 13 figures, conference paper

Large Language Models (LLMs) pre-trained on internet-scale datasets have shown impressive capabilities in code understanding, synthesis, and general purpose question-and-answering. Key to their performance is the substantial prior knowledge acquired during training and their ability to reason over extended sequences of symbols, often presented in natural language. In this work, we aim to harness the extensive long-term reasoning, natural language comprehension, and the available prior knowledge of LLMs for increased resilience and adaptation in autonomous mobile robots. We introduce REAL, an approach for REsilience and Adaptation using LLMs. REAL provides a strategy to employ LLMs as a part of the mission planning and control framework of an autonomous robot. The LLM employed by REAL provides (i) a source of prior knowledge to increase resilience for challenging scenarios that the system had not been explicitly designed for; (ii) a way to interpret natural-language and other log/diagnostic information available in the autonomy stack, for mission planning; (iii) a way to adapt the control inputs using minimal user-provided prior knowledge about the dynamics/kinematics of the robot. We integrate REAL in the autonomy stack of a real multirotor, querying onboard an offboard LLM at 0.1-1.0 Hz as part the robot's mission planning and control feedback loops. We demonstrate in real-world experiments the ability of the LLM to reduce the position tracking errors of a multirotor under the presence of (i) errors in the parameters of the controller and (ii) unmodeled dynamics. We also show (iii) decision making to avoid potentially dangerous scenarios (e.g., robot oscillates) that had not been explicitly accounted for in the initial prompt design., Comment: 13 pages, 5 figures, conference workshop

We describe how the nested soft-collinear subtraction scheme [1] can be used to compute the next-to-next-to-leading order (NNLO) QCD corrections to the production of an arbitrary number of gluonic jets in hadron collisions. We show that the infrared subtraction terms can be combined into recurring structures that in many cases are simple iterations of those terms known from next-to-leading order. The way that these recurring structures are identified and computed is fairly general, and can be applied to any partonic process. As an example, we explicitly demonstrate the cancellation of all singularities in the fully-differential cross section for the $q\bar{q} \to X + N g $ process at NNLO in QCD. The finite remainder of the NNLO QCD contribution, which arises upon cancellation of all $\epsilon$-poles, is expressed via relatively simple formulas, which can be implemented in a numerical code in a straightforward way. Our approach can be extended to describe arbitrary processes at NNLO in QCD; the largest remaining challenge at this point is the combinatorics of quark and gluon collinear limits., Comment: 116 pages, 9 appendices, 1 figure, alphabetic index

Large Language Models (LLMs) are deployed in interactive contexts with direct user engagement, such as chatbots and writing assistants. These deployments are vulnerable to prompt injection and jailbreaking (collectively, prompt hacking), in which models are manipulated to ignore their original instructions and follow potentially malicious ones. Although widely acknowledged as a significant security threat, there is a dearth of large-scale resources and quantitative studies on prompt hacking. To address this lacuna, we launch a global prompt hacking competition, which allows for free-form human input attacks. We elicit 600K+ adversarial prompts against three state-of-the-art LLMs. We describe the dataset, which empirically verifies that current LLMs can indeed be manipulated via prompt hacking. We also present a comprehensive taxonomical ontology of the types of adversarial prompts., Comment: 34 pages, 8 figures Codebase: https://github.com/PromptLabs/hackaprompt Dataset: https://huggingface.co/datasets/hackaprompt/hackaprompt-dataset/blob/main/README.md Playground: https://huggingface.co/spaces/hackaprompt/playground

CdTe is the most commercially successful thin-film photovoltaic technology to date. The recent development of Se-alloyed CdSe$_{x}$Te$_{1-x}$ layers in CdTe solar cells has led to higher device efficiencies, due to a lowered bandgap improving the photocurrent, improved voltage characteristics and longer carrier lifetimes. Evidence from cross-sectional electron microscopy is widely believed to indicate that Se passivates defects in CdSe$_{x}$Te$_{1-x}$ solar cells, and that this is the reason for better lifetimes and voltages in these devices. Here, we utilise spatially resolved photoluminescence measurements of CdSe$_{x}$Te$_{1-x}$ thin films on glass to study the effects of Se on carrier recombination in the material, isolated from the impact of conductive interfaces and without the need to prepare cross-sections through the samples. We find further evidence to support Se passivation of grain boundaries, but also identify an associated increase in below-bandgap photoluminescence that indicates the presence of Se-enhanced luminescent defects. Our results show that Se treatment, in tandem with Cl passivation, does increase radiative efficiencies. However, the simultaneous enhancement of defects within the grain interiors suggests that although it is overall beneficial, Se incorporation may still ultimately limit the maximum attainable efficiency of CdSe$_{x}$Te$_{1-x}$ solar cells.

Evaporation-driven fluid flow in porous or nanostructured materials has recently opened a new paradigm for renewable energy generation. Despite recent progress, major fundamental questions remain regarding the interfacial phenomena governing these so-called hydrovoltaic (HV) devices. Together with the lack of modelling tools, this limits the performance and application range of this emerging technology. By leveraging ordered arrays of Silicon nanopillars (NP) and developing a quantitative multiphysics model to study their HV response across a wide parameter space, this work reveals the complex interplay of surface-charge, liquid properties, and geometrical parameters, including previously unexplored electrokinetic interactions. Notably, we find that ion-concentration-dependent surface charge, together with ion mobility, dictates multiple local maxima in open circuit voltage, with optimal conditions deviating from conventional low-concentration expectations. Additionally, assessing the HV response up to molar concentrations, we provide unique evidence of ion adsorption and charge inversion for a number of monovalent cations. This effect interestingly enables the operation of HV devices even at such high concentrations. Finally, we highlight that, beyond electrokinetic parameters, geometrical asymmetries in the device structure generate an electrostatic potential that augments HV performance. Overall, our work, which lies in between single nanochannel studies and macro-scale porous system characterization, demonstrates that evaporation-driven HV devices can operate across a wide range of salinities, with optimal operating conditions being dictated by distinct interfacial phenomena. Thus it offers crucial insight and a design tool for enhancing the performance of evaporation-driven HV devices and enables their broader applicability across the salinity scale of natural and processed waters.

We discuss a path toward the generalisation of the nested soft-collinear subtraction scheme to arbitrary $2\rightarrow n$ processes. The scheme is designed to provide an efficient and process-independent procedure to extract and regulate infrared (IR) singularities arising from unresolved real radiation and combine them with explicit singularities in virtual corrections. The new approach is based on a reorganisation of the relevant subtraction terms into simple combinations of a relatively small number of recurring structures. This strategy leads to a drastic reduction in the computational effort required to derive integrated subtraction terms, while preserving the full generality of the scheme. We believe that this approach will allow for tackling the issue of regularising IR divergences at next-to-next-to-leading order in the strong coupling constant for arbitrary, multi-parton processes.

The recently proposed Data Lakehouse architecture is built on open file formats, performance, and first-class support for data transformation, BI and data science: while the vision stresses the importance of lowering the barrier for data work, existing implementations often struggle to live up to user expectations. At Bauplan, we decided to build a new serverless platform to fulfill the Lakehouse vision. Since building from scratch is a challenge unfit for a startup, we started by re-using (sometimes unconventionally) existing projects, and then investing in improving the areas that would give us the highest marginal gains for the developer experience. In this work, we review user experience, high-level architecture and tooling decisions, and conclude by sharing plans for future development., Comment: Paper accepted for the Second International Workshop on Composable Data Management Systems (@ VLDB 2023)

Harnessing non-equilibrium hot carriers from plasmonic metal nanostructures constitutes a vibrant research field. It promises to enable control of activity and selectivity of photochemical reactions, especially for solar fuel generation. However, a comprehensive understanding of the interplay of plasmonic hot carrier-driven processes in metal/semiconducting heterostructures has remained elusive. In this work, we reveal the complex interdependence between plasmon excitation, hot carrier generation, transport and interfacial collection in plasmonic photocatalytic devices, uniquely determining the charge injection efficiencies at the solid/solid and solid/liquid interfaces. Interestingly, by measuring the internal quantum efficiency of ultrathin (14 to 33 nm) single-crystalline plasmonic gold (Au) nanoantenna arrays on titanium dioxide substrates, we find that the performance of the device is governed by hot hole collection at the metal/electrolyte interface. In particular, by combining a solid- and liquid-state experimental approach with ab initio simulations, we show a more efficient collection of high-energy d-band holes traveling in [111] orientation, resulting in a stronger oxidation reaction at the {111} surfaces of the nanoantenna. These results thus establish new guidelines for the design and optimization of plasmonic photocatalytic systems and optoelectronic devices.

Luminescence constitutes a unique source of insight into hot carrier processes in metals, including those in plasmonic nanostructures used for sensing and energy applications. However, being weak in nature, metal luminescence remains poorly understood, its microscopic origin strongly debated, and its potential for unravelling nanoscale carrier dynamics largely unexploited. Here, we reveal quantum-mechanical effects emanating in the luminescence from thin monocrystalline gold flakes. Specifically, we present experimental evidence, supported by first-principles simulations, to demonstrate its photoluminescence origin when exciting in the interband regime. Our model allows us to identify changes to the measured gold luminescence due to quantum-mechanical effects as the gold film thickness is reduced. Excitingly, such effects are observable in the luminescence signal from flakes up to 40 nm in thickness, associated with the out-of-plane discreteness of the electronic band structure near the Fermi level. We qualitatively reproduce the observations with first-principles modelling, thus establishing a unified description of luminescence in gold and enabling its widespread application as a probe of carrier dynamics and light-matter interactions in this material. Our study paves the way for future explorations of hot-carriers and charge-transfer dynamics in a multitude of material systems., Comment: Main text 21 pages and 4 figures. Supplemental Information 33 pages and 17 figures

Imitation Learning (IL) can generate computationally efficient policies from demonstrations provided by Model Predictive Control (MPC). However, IL methods often require extensive data-collection and training efforts, limiting changes to the policy if the task changes, and they produce policies with limited robustness to new disturbances. In this work, we propose an IL strategy to efficiently compress a computationally expensive MPC into a deep neural network policy that is robust to previously unseen disturbances. By using a robust variant of the MPC, called Robust Tube MPC, and leveraging properties from the controller, we introduce computationally efficient data augmentation methods that enable a significant reduction of the number of MPC demonstrations and training efforts required to generate a robust policy. Our approach opens the possibility of zero-shot transfer of a policy trained from a single MPC demonstration collected in a nominal domain, such as a simulation or a robot in a lab/controlled environment, to a new domain with previously unseen bounded model errors/perturbations. Numerical evaluations performed using linear and nonlinear MPC for agile flight on a multirotor show that our method outperforms strategies commonly employed in IL (such as Dataset-Aggregation (DAgger) and Domain Randomization (DR)) in terms of demonstration-efficiency, training time, and robustness to perturbations unseen during training. Experimental evaluations validate the efficiency and real-world robustness., Comment: Accepted for the Transaction on Robotics (T-RO), May 2024

As ecommerce continues growing, huge investments in ML and NLP for Information Retrieval are following. While the vector space model dominated retrieval modelling in product search - even as vectorization itself greatly changed with the advent of deep learning -, our position paper argues in a contrarian fashion that program synthesis provides significant advantages for many queries and a significant number of players in the market. We detail the industry significance of the proposed approach, sketch implementation details, and address common objections drawing from our experience building a similar system at Tooso., Comment: Published at SIGIR eCom 2023

Recommender Systems today are still mostly evaluated in terms of accuracy, with other aspects beyond the immediate relevance of recommendations, such as diversity, long-term user retention and fairness, often taking a back seat. Moreover, reconciling multiple performance perspectives is by definition indeterminate, presenting a stumbling block to those in the pursuit of rounded evaluation of Recommender Systems. EvalRS 2022 -- a data challenge designed around Multi-Objective Evaluation -- was a first practical endeavour, providing many insights into the requirements and challenges of balancing multiple objectives in evaluation. In this work, we reflect on EvalRS 2022 and expound upon crucial learnings to formulate a first-principles approach toward Multi-Objective model selection, and outline a set of guidelines for carrying out a Multi-Objective Evaluation challenge, with potential applicability to the problem of rounded evaluation of competing models in real-world deployments., Comment: 15 pages, under submission

EvalRS aims to bring together practitioners from industry and academia to foster a debate on rounded evaluation of recommender systems, with a focus on real-world impact across a multitude of deployment scenarios. Recommender systems are often evaluated only through accuracy metrics, which fall short of fully characterizing their generalization capabilities and miss important aspects, such as fairness, bias, usefulness, informativeness. This workshop builds on the success of last year's workshop at CIKM, but with a broader scope and an interactive format., Comment: EvalRS 2023 is a workshop at KDD23. Code and hackathon materials: https://github.com/RecList/evalRS-KDD-2023

The deployment of agile autonomous systems in challenging, unstructured environments requires adaptation capabilities and robustness to uncertainties. Existing robust and adaptive controllers, such as those based on model predictive control (MPC), can achieve impressive performance at the cost of heavy online onboard computations. Strategies that efficiently learn robust and onboard-deployable policies from MPC have emerged, but they still lack fundamental adaptation capabilities. In this work, we extend an existing efficient Imitation Learning (IL) algorithm for robust policy learning from MPC with the ability to learn policies that adapt to challenging model/environment uncertainties. The key idea of our approach consists in modifying the IL procedure by conditioning the policy on a learned lower-dimensional model/environment representation that can be efficiently estimated online. We tailor our approach to the task of learning an adaptive position and attitude control policy to track trajectories under challenging disturbances on a multirotor. Evaluations in simulation show that a high-quality adaptive policy can be obtained in about $1.3$ hours. We additionally empirically demonstrate rapid adaptation to in- and out-of-training-distribution uncertainties, achieving a $6.1$ cm average position error under wind disturbances that correspond to about $50\%$ of the weight of the robot, and that are $36\%$ larger than the maximum wind seen during training., Comment: 8 pages, 6 figures. IROS23 final version, with page numbers added

As Machine Learning (ML) gains adoption across industries and new use cases, practitioners increasingly realize the challenges around effectively developing and iterating on ML systems: reproducibility, debugging, scalability, and documentation are elusive goals for real-world pipelines outside tech-first companies. In this paper, we review the nature of ML-oriented workloads and argue that re-purposing existing tools won't solve the current productivity issues, as ML peculiarities warrant specialized development tooling. We then introduce Metaflow, an open-source framework for ML projects explicitly designed to boost the productivity of data practitioners by abstracting away the execution of ML code from the definition of the business logic. We show how our design addresses the main challenges in ML operations (MLOps), and document through examples, interviews and use cases its practical impact on the field.