Search

Frontier models are increasingly trained and deployed as autonomous agent. One safety concern is that AI agents might covertly pursue misaligned goals, hiding their true capabilities and objectives - also known as scheming. We study whether models have the capability to scheme in pursuit of a goal that we provide in-context and instruct the model to strongly follow. We evaluate frontier models on a suite of six agentic evaluations where models are instructed to pursue goals and are placed in environments that incentivize scheming. Our results show that o1, Claude 3.5 Sonnet, Claude 3 Opus, Gemini 1.5 Pro, and Llama 3.1 405B all demonstrate in-context scheming capabilities. They recognize scheming as a viable strategy and readily engage in such behavior. For example, models strategically introduce subtle mistakes into their responses, attempt to disable their oversight mechanisms, and even exfiltrate what they believe to be their model weights to external servers. Additionally, this deceptive behavior proves persistent. When o1 has engaged in scheming, it maintains its deception in over 85% of follow-up questions and often remains deceptive in multi-turn interrogations. Analysis of the models' chains-of-thought reveals that models explicitly reason about these deceptive strategies, providing evidence that the scheming behavior is not accidental. Surprisingly, we also find rare instances where models engage in scheming when only given a goal, without being strongly nudged to pursue it. We observe cases where Claude 3.5 Sonnet strategically underperforms in evaluations in pursuit of being helpful, a goal that was acquired during training rather than in-context. Our findings demonstrate that frontier models now possess capabilities for basic in-context scheming, making the potential of AI agents to engage in scheming behavior a concrete rather than theoretical concern.

Assistive technology can be leveraged by blind people when searching for objects in their daily lives. We created ObjectFinder, an open-vocabulary interactive object-search prototype, which combines object detection with scene description and navigation. It enables blind persons to detect and navigate to objects of their choice. Our approach used co-design for the development of the prototype. We further conducted need-finding interviews to better understand challenges in object search, followed by a study with the ObjectFinder prototype in a laboratory setting simulating a living room and an office, with eight blind users. Additionally, we compared the prototype with BeMyEyes and Lookout for object search. We found that most participants felt more independent with ObjectFinder and preferred it over the baselines when deployed on more efficient hardware, as it enhances mental mapping and allows for active target definition. Moreover, we identified factors for future directions for the development of object-search systems.

Monocular geometric scene understanding combines panoptic segmentation and self-supervised depth estimation, focusing on real-time application in autonomous vehicles. We introduce MGNiceNet, a unified approach that uses a linked kernel formulation for panoptic segmentation and self-supervised depth estimation. MGNiceNet is based on the state-of-the-art real-time panoptic segmentation method RT-K-Net and extends the architecture to cover both panoptic segmentation and self-supervised monocular depth estimation. To this end, we introduce a tightly coupled self-supervised depth estimation predictor that explicitly uses information from the panoptic path for depth prediction. Furthermore, we introduce a panoptic-guided motion masking method to improve depth estimation without relying on video panoptic segmentation annotations. We evaluate our method on two popular autonomous driving datasets, Cityscapes and KITTI. Our model shows state-of-the-art results compared to other real-time methods and closes the gap to computationally more demanding methods. Source code and trained models are available at https://github.com/markusschoen/MGNiceNet., Comment: Accepted for ACCV 2024

Depth estimation is an essential task toward full scene understanding since it allows the projection of rich semantic information captured by cameras into 3D space. While the field has gained much attention recently, datasets for depth estimation lack scene diversity or sensor modalities. This work presents the ADUULM-360 dataset, a novel multi-modal dataset for depth estimation. The ADUULM-360 dataset covers all established autonomous driving sensor modalities, cameras, lidars, and radars. It covers a frontal-facing stereo setup, six surround cameras covering the full 360-degree, two high-resolution long-range lidar sensors, and five long-range radar sensors. It is also the first depth estimation dataset that contains diverse scenes in good and adverse weather conditions. We conduct extensive experiments using state-of-the-art self-supervised depth estimation methods under different training tasks, such as monocular training, stereo training, and full surround training. Discussing these results, we demonstrate common limitations of state-of-the-art methods, especially in adverse weather conditions, which hopefully will inspire future research in this area. Our dataset, development kit, and trained baselines are available at https://github.com/uulm-mrm/aduulm_360_dataset., Comment: 2024 IEEE International Conference on Intelligent Transportation Systems (ITSC)

Important advances have recently been made in the search for materials with complex multi-phase landscapes that host photoinduced metastable collective states with exotic functionalities. In almost all cases so far, the desired phases are accessed by exploiting light-matter interactions via the imaginary part of the dielectric function through above-bandgap or resonant mode excitation. Nonresonant Raman excitation of coherent modes has been experimentally observed and proposed for dynamic material control, but the resulting atomic excursion has been limited to perturbative levels. Here, we demonstrate that it is possible to overcome this challenge by employing nonresonant ultrashort pulses with low photon energies well below the bandgap. Using mid-infrared pulses, we induce ferroelectric reversal in lithium niobate and phase switching in tin selenide and characterize the large-amplitude mode displacements through femtosecond Raman scattering, second harmonic generation, and x-ray diffraction. This approach, validated by first-principle calculations, defines a novel method for synthesizing hidden phases with unique functional properties and manipulating complex energy landscapes at reduced energy consumption and ultrafast speeds., Comment: 5 figures

With the recent advancements in Large Language Models (LLMs), web developers increasingly apply their code-generation capabilities to website design. However, since these models are trained on existing designerly knowledge, they may inadvertently replicate bad or even illegal practices, especially deceptive designs (DD). This paper examines whether users can accidentally create DD for a fictitious webshop using GPT-4. We recruited 20 participants, asking them to use ChatGPT to generate functionalities (product overview or checkout) and then modify these using neutral prompts to meet a business goal (e.g., "increase the likelihood of us selling our product"). We found that all 20 generated websites contained at least one DD pattern (mean: 5, max: 9), with GPT-4 providing no warnings. When reflecting on the designs, only 4 participants expressed concerns, while most considered the outcomes satisfactory and not morally problematic, despite the potential ethical and legal implications for end-users and those adopting ChatGPT's recommendations

For the validation and verification of automotive radars, datasets of realistic traffic scenarios are required, which, how ever, are laborious to acquire. In this paper, we introduce radar scene synthesis using GANs as an alternative to the real dataset acquisition and simulation-based approaches. We train a PointNet++ based GAN model to generate realistic radar point cloud scenes and use a binary classifier to evaluate the performance of scenes generated using this model against a test set of real scenes. We demonstrate that our GAN model achieves similar performance (~87%) to the real scenes test set., Comment: ICMIM 2024; 7th IEEE MTT Conference

Adversarial training can be used to learn models that are robust against perturbations. For linear models, it can be formulated as a convex optimization problem. Compared to methods proposed in the context of deep learning, leveraging the optimization structure allows significantly faster convergence rates. Still, the use of generic convex solvers can be inefficient for large-scale problems. Here, we propose tailored optimization algorithms for the adversarial training of linear models, which render large-scale regression and classification problems more tractable. For regression problems, we propose a family of solvers based on iterative ridge regression and, for classification, a family of solvers based on projected gradient descent. The methods are based on extended variable reformulations of the original problem. We illustrate their efficiency in numerical examples.

Image monitoring and guidance during medical examinations can aid both diagnosis and treatment. However, the sampling frequency is often too low, which creates a need to estimate the missing images. We present a probabilistic motion model for sequential medical images, with the ability to both estimate motion between acquired images and forecast the motion ahead of time. The core is a low-dimensional temporal process based on a linear Gaussian state-space model with analytically tractable solutions for forecasting, simulation, and imputation of missing samples. The results, from two experiments on publicly available cardiac datasets, show reliable motion estimates and an improved forecasting performance using patient-specific adaptation by online learning., Comment: Medical Image Computing and Computer Assisted Intervention (MICCAI) 2024

Since neural networks can make wrong predictions even with high confidence, monitoring their behavior at runtime is important, especially in safety-critical domains like autonomous driving. In this paper, we combine ideas from previous monitoring approaches based on observing the activation values of hidden neurons. In particular, we combine the Gaussian-based approach, which observes whether the current value of each monitored neuron is similar to typical values observed during training, and the Outside-the-Box monitor, which creates clusters of the acceptable activation values, and, thus, considers the correlations of the neurons' values. Our experiments evaluate the achieved improvement.

Capturing images using multispectral camera arrays has gained importance in medical, agricultural and environmental processes. However, using all available spectral bands is infeasible and produces much data, while only a fraction is needed for a given task. Nearby bands may contain similar information, therefore redundant spectral bands should not be considered in the evaluation process to keep complexity and the data load low. In current methods, a restricted and pre-determined number of spectral bands is selected. Our approach improves this procedure by including preset conditions such as noise or the bandwidth of available filters, minimizing spectral redundancy. Furthermore, a minimal filter selection can be conducted, keeping the hardware setup at low costs, while still obtaining all important spectral information. In comparison to the fast binary search filter band selection method, we managed to reduce the amount of misclassified objects of the SMM dataset from 318 to 124 using a random forest classifier.

For run sizes that are a multiple of four, the literature offers many two-level designs that are D- and A-optimal for the main-effects model and minimize the aliasing between main effects and interaction effects and among interaction effects. For run sizes that are not a multiple of four, no conclusive results are known. In this paper, we propose two algorithms that generate all non-isomorphic D- and A-optimal main-effects designs for run sizes that are one and two more than a multiple of four. We enumerate all such designs for run sizes up to 18, report the numbers of designs we obtained, and identify those that minimize the aliasing between main effects and interaction effects and among interaction effects. Finally, we compare the minimally aliased designs we found with benchmark designs from the literature.

Diffusion models have achieved remarkable progress in generative modelling, particularly in enhancing image quality to conform to human preferences. Recently, these models have also been applied to low-level computer vision for photo-realistic image restoration (IR) in tasks such as image denoising, deblurring, dehazing, etc. In this review paper, we introduce key constructions in diffusion models and survey contemporary techniques that make use of diffusion models in solving general IR tasks. Furthermore, we point out the main challenges and limitations of existing diffusion-based IR frameworks and provide potential directions for future work., Comment: Review paper; any comments and suggestions are most welcome!

Generative diffusions are a powerful class of Monte Carlo samplers that leverage bridging Markov processes to approximate complex, high-dimensional distributions, such as those found in image processing and language models. Despite their success in these domains, an important open challenge remains: extending these techniques to sample from conditional distributions, as required in, for example, Bayesian inverse problems. In this paper, we present a comprehensive review of existing computational approaches to conditional sampling within generative diffusion models. Specifically, we highlight key methodologies that either utilise the joint distribution, or rely on (pre-trained) marginal distributions with explicit likelihoods, to construct conditional generative samplers.

Uncertainty estimation is a necessary component when implementing AI in high-risk settings, such as autonomous cars, medicine, or insurances. Large Language Models (LLMs) have seen a surge in popularity in recent years, but they are subject to hallucinations, which may cause serious harm in high-risk settings. Despite their success, LLMs are expensive to train and run: they need a large amount of computations and memory, preventing the use of ensembling methods in practice. In this work, we present a novel method that allows for fast and memory-friendly training of LLM ensembles. We show that the resulting ensembles can detect hallucinations and are a viable approach in practice as only one GPU is needed for training and inference., Comment: 6 pages, 3 figures

Safe Bayesian optimization (BO) algorithms promise to find optimal control policies without knowing the system dynamics while at the same time guaranteeing safety with high probability. In exchange for those guarantees, popular algorithms require a smoothness assumption: a known upper bound on a norm in a reproducing kernel Hilbert space (RKHS). The RKHS is a potentially infinite-dimensional space, and it is unclear how to, in practice, obtain an upper bound of an unknown function in its corresponding RKHS. In response, we propose an algorithm that estimates an upper bound on the RKHS norm of an unknown function from data and investigate its theoretical properties. Moreover, akin to Lipschitz-based methods, we treat the RKHS norm as a local rather than a global object, and thus reduce conservatism. Integrating the RKHS norm estimation and the local interpretation of the RKHS norm into a safe BO algorithm yields PACSBO, an algorithm for probably approximately correct safe Bayesian optimization, for which we provide numerical and hardware experiments that demonstrate its applicability and benefits over popular safe BO algorithms., Comment: Accepted to the Symposium on Systems Theory in Data and Optimization (SysDO 2024). This is a preprint of the final version, which is to appear in Lecture Notes in Control and Information Sciences - Proceedings

In this report we investigate the use of the Tustin neural network architecture (Tustin-Net) for the identification of a physical rotary inverse pendulum. This physics-based architecture is of particular interest as it builds on the known relationship between velocities and positions. We here aim at discussing the advantages, limitations and performance of Tustin-Nets compared to first-principles grey-box models on a real physical apparatus, showing how, with a standard training procedure, the former can hardly achieve the same accuracy as the latter. To address this limitation, we present a training strategy based on transfer learning that yields Tustin-Nets that are competitive with the first-principles model, without requiring extensive knowledge of the setup as the latter.

To advance formal verification of stochastic systems against temporal logic requirements for handling unknown dynamics, researchers have been designing data-driven approaches inspired by breakthroughs in the underlying machine learning techniques. As one promising research direction, abstraction-based solutions based on Gaussian process (GP) regression have become popular for their ability to learn a representation of the latent system from data with a quantified error. Results obtained based on this model are then translated to the true system via various methods. In a recent publication, GPs using a so-called binary-tree kernel have demonstrated a polynomial speedup w.r.t. the size of the data compared to their vanilla version, outcompeting all existing sparse GP approximations. Incidentally, the resulting binary-tree Gaussian process (BTGP) is characteristic for its piecewise-constant posterior mean and covariance functions, naturally abstracting the input space into discrete partitions. In this paper, we leverage this natural abstraction of the BTGP for formal verification, eliminating the need for cumbersome abstraction and error quantification procedures. We show that the BTGP allows us to construct an interval Markov chain model of the unknown system with a speedup that is polynomial w.r.t. the size of the abstraction compared to alternative approaches. We provide a delocalized error quantification via a unified formula even when the true dynamics do not live in the function space of the BTGP. This allows us to compute upper and lower bounds on the probability of satisfying reachability specifications that are robust to both aleatoric and epistemic uncertainties., Comment: Published at IFAC conference on analysis and design of hybrid systems (ADHS) 2024

In this paper we introduce a new dataset containing instance segmentation masks for ten different categories of winter sports equipment, called WSESeg (Winter Sports Equipment Segmentation). Furthermore, we carry out interactive segmentation experiments on said dataset to explore possibilities for efficient further labeling. The SAM and HQ-SAM models are conceptualized as foundation models for performing user guided segmentation. In order to measure their claimed generalization capability we evaluate them on WSESeg. Since interactive segmentation offers the benefit of creating easily exploitable ground truth data during test-time, we are going to test various online adaptation methods for the purpose of exploring potentials for improvements without having to fine-tune the models explicitly. Our experiments show that our adaptation methods drastically reduce the Failure Rate (FR) and Number of Clicks (NoC) metrics, which generally leads faster to better interactive segmentation results., Comment: 7 pages, 1 figure, 3 tables, Accepted at CBMI 2024

In many manufacturing settings, annotating data for machine learning and computer vision is costly, but synthetic data can be generated at significantly lower cost. Substituting the real-world data with synthetic data is therefore appealing for many machine learning applications that require large amounts of training data. However, relying solely on synthetic data is frequently inadequate for effectively training models that perform well on real-world data, primarily due to domain shifts between the synthetic and real-world data. We discuss approaches for dealing with such a domain shift when detecting defects in X-ray scans of aluminium wheels. Using both simulated and real-world X-ray images, we train an object detection model with different strategies to identify the training approach that generates the best detection results while minimising the demand for annotated real-world training samples. Our preliminary findings suggest that the sim-2-real domain adaptation approach is more cost-efficient than a fully supervised oracle - if the total number of available annotated samples is fixed. Given a certain number of labeled real-world samples, training on a mix of synthetic and unlabeled real-world data achieved comparable or even better detection results at significantly lower cost. We argue that future research into the cost-efficiency of different training strategies is important for a better understanding of how to allocate budget in applied machine learning projects.

This paper presents the development of a lexicon centered on emerging concepts, focusing on non-technological innovation. It introduces a four-step methodology that combines human expertise, statistical analysis, and machine learning techniques to establish a model that can be generalized across multiple domains. This process includes the creation of a thematic corpus, the development of a Gold Standard Lexicon, annotation and preparation of a training corpus, and finally, the implementation of learning models to identify new terms. The results demonstrate the robustness and relevance of our approach, highlighting its adaptability to various contexts and its contribution to lexical research. The developed methodology promises applicability in conceptual fields., Comment: in French language. JADT 2024

Context: In agile transformations, there are many challenges such as alignment between agile practices and the organizational goals and strategies or issues with shifts in how work is organized and executed. One very important challenge but less considered and treated in research are cultural challenges associated with an agile mindset. Although research shows that cultural clashes and general organizational resistance to change are part of the most significant agile adoption barriers. Objective: We identify challenges that arise from the interplay between agile culture and organizational culture. In doing so, we tackle this field and come up with important contributions for further research regarding a problem that practitioners face today. Method: This is done with a mixed-method research approach. First, we gathered qualitative data among our network of agile practitioners and derived in sum 15 challenges with agile culture. Then, we conducted quantitative data by means of a questionnaire study with 92 participants. Results: We identified 7 key challenges out of the 15 challenges with agile culture. These key challenges refer to the technical agility (doing agile) and the cultural agility (being agile). The results are presented in type of a conceptual model named the Agile Cultural Challenges (ACuCa). Conclusion: Based on our results, we started deriving future work aspects to do more detailed research on the topic of cultural challenges while transitioning or using agile methods in software development and beyond., Comment: arXiv admin note: text overlap with arXiv:2212.07218

Utilizing transformer architectures for semantic segmentation of high-resolution images is hindered by the attention's quadratic computational complexity in the number of tokens. A solution to this challenge involves decreasing the number of tokens through token merging, which has exhibited remarkable enhancements in inference speed, training efficiency, and memory utilization for image classification tasks. In this paper, we explore various token merging strategies within the framework of the Segformer architecture and perform experiments on multiple semantic segmentation and human pose estimation datasets. Notably, without model re-training, we, for example, achieve an inference acceleration of 61% on the Cityscapes dataset while maintaining the mIoU performance. Consequently, this paper facilitates the deployment of transformer-based architectures on resource-constrained devices and in real-time applications., Comment: 7 pages, to be published in IEEE International Conference on Multimedia Information Processing and Retrieval (MIPR) 2024

Given an unconditional diffusion model $\pi(x, y)$, using it to perform conditional simulation $\pi(x \mid y)$ is still largely an open question and is typically achieved by learning conditional drifts to the denoising SDE after the fact. In this work, we express conditional simulation as an inference problem on an augmented space corresponding to a partial SDE bridge. This perspective allows us to implement efficient and principled particle Gibbs and pseudo-marginal samplers marginally targeting the conditional distribution $\pi(x \mid y)$. Contrary to existing methodology, our methods do not introduce any additional approximation to the unconditional diffusion model aside from the Monte Carlo error. We showcase the benefits and drawbacks of our approach on a series of synthetic and real data examples., Comment: 24 pages, 12 figures

Cold atom interferometry (CAI)-based quantum accelerometers are very promising for future satellite gravity missions thanks to their strength in providing long-term stable and precise measurements of non-gravitational accelerations. However, their limitations due to the low measurement rate and the existence of ambiguities in the raw sensor measurements call for hybridization of the quantum accelerometer (Q-ACC) with a classical one (e.g., electrostatic) with higher bandwidth. While previous hybridization studies have so far considered simple noise models for the Q-ACC and neglected the impact of satellite rotation on the phase shift of the accelerometer, we perform here a more advanced hybridization simulation by implementing a comprehensive noise model for the satellite-based quantum accelerometers and considering the full impact of rotation, gravity gradient, and self-gravity on the instrument. We perform simulation studies for scenarios with different assumptions about quantum and classical sensors and satellite missions. The performance benefits of the hybrid solutions, taking the synergy of both classical and quantum accelerometers into account, will be quantified. We found that implementing a hybrid accelerometer onboard a future gravity mission improves the gravity solution by one to two orders in lower and higher degrees. In particular, the produced global gravity field maps show a drastic reduction in the instrumental contribution to the striping effect after introducing measurements from the hybrid accelerometers.

We define a framework for incorporating alternation-free fixpoint logics into the dual-adjunction setup for coalgebraic modal logics. We achieve this by using order-enriched categories. We give a least-solution semantics as well as an initial algebra semantics, and prove they are equivalent. We also show how to place the alternation-free coalgebraic $\mu$-calculus in this framework, as well as PDL and a logic with a probabilistic dynamic modality.

Recent advances in cold atom interferometry have cleared the path for space applications of quantum inertial sensors, whose level of stability is expected to increase dramatically with the longer interrogation times accessible in space. In this study, a comprehensive in-orbit model is developed for a Mach-Zehnder-type cold-atom accelerometer. Performance tests are realized under different assumptions, and the impact of various sources of errors on instrument stability is evaluated. Current and future advances for space-based atom interferometry are discussed, and their impact on the performance of quantum sensors on-board satellite gravity missions is investigated in three different scenarios: state-of-the-art scenario, near-future (between the next 5 and 10 years) and far-future scenarios (between the next 10 to 20 years). We show that one can achieve a sensitivity level close to 5E-10 with the current state-of-the-art technology. We also estimate that in the near and far-future, atom interferometry in space is expected to achieve sensitivity levels of 1E-11 and 1E-12, respectively. A roadmap for improvements in atom interferometry is provided that would maximize the performance of future CAI accelerometers, considering their technical capabilities. Finally, the possibility and challenges of having ultra-sensitive atom interferometry in space for future space missions are discussed., Comment: Priprint of the article submitted to the JASR, 17 Pages

Though diffusion models have been successfully applied to various image restoration (IR) tasks, their performance is sensitive to the choice of training datasets. Typically, diffusion models trained in specific datasets fail to recover images that have out-of-distribution degradations. To address this problem, this work leverages a capable vision-language model and a synthetic degradation pipeline to learn image restoration in the wild (wild IR). More specifically, all low-quality images are simulated with a synthetic degradation pipeline that contains multiple common degradations such as blur, resize, noise, and JPEG compression. Then we introduce robust training for a degradation-aware CLIP model to extract enriched image content features to assist high-quality image restoration. Our base diffusion model is the image restoration SDE (IR-SDE). Built upon it, we further present a posterior sampling strategy for fast noise-free image generation. We evaluate our model on both synthetic and real-world degradation datasets. Moreover, experiments on the unified image restoration task illustrate that the proposed posterior sampling improves image generation quality for various degradations., Comment: CVPRW 2024; Code: https://github.com/Algolzw/daclip-uir

Scene graph generation has emerged as a prominent research field in computer vision, witnessing significant advancements in the recent years. However, despite these strides, precise and thorough definitions for the metrics used to evaluate scene graph generation models are lacking. In this paper, we address this gap in the literature by providing a review and precise definition of commonly used metrics in scene graph generation. Our comprehensive examination clarifies the underlying principles of these metrics and can serve as a reference or introduction to scene graph metrics. Furthermore, to facilitate the usage of these metrics, we introduce a standalone Python package called SGBench that efficiently implements all defined metrics, ensuring their accessibility to the research community. Additionally, we present a scene graph benchmarking web service, that enables researchers to compare scene graph generation methods and increase visibility of new methods in a central place. All of our code can be found at https://lorjul.github.io/sgbench/.

The interactive segmentation task consists in the creation of object segmentation masks based on user interactions. The most common way to guide a model towards producing a correct segmentation consists in clicks on the object and background. The recently published Segment Anything Model (SAM) supports a generalized version of the interactive segmentation problem and has been trained on an object segmentation dataset which contains 1.1B masks. Though being trained extensively and with the explicit purpose of serving as a foundation model, we show significant limitations of SAM when being applied for interactive segmentation on novel domains or object types. On the used datasets, SAM displays a failure rate $\text{FR}_{30}@90$ of up to $72.6 \%$. Since we still want such foundation models to be immediately applicable, we present a framework that can adapt SAM during immediate usage. For this we will leverage the user interactions and masks, which are constructed during the interactive segmentation process. We use this information to generate pseudo-labels, which we use to compute a loss function and optimize a part of the SAM model. The presented method causes a relative reduction of up to $48.1 \%$ in the $\text{FR}_{20}@85$ and $46.6 \%$ in the $\text{FR}_{30}@90$ metrics., Comment: 11 pages, 2 figures, 4 tables

Algorithmic verification of realistic systems to satisfy safety and other temporal requirements has suffered from poor scalability of the employed formal approaches. To design systems with rigorous guarantees, many approaches still rely on exact models of the underlying systems. Since this assumption can rarely be met in practice, models have to be inferred from measurement data or are bypassed completely. Whilst former usually requires the model structure to be known a-priori and immense amounts of data to be available, latter gives rise to a plethora of restrictive mathematical assumptions about the unknown dynamics. In a pursuit of developing scalable formal verification algorithms without shifting the problem to unrealistic assumptions, we employ the concept of barrier certificates, which can guarantee safety of the system, and learn the certificate directly from a compact set of system trajectories. We use conditional mean embeddings to embed data from the system into a reproducing kernel Hilbert space (RKHS) and construct an RKHS ambiguity set that can be inflated to robustify the result w.r.t. a set of plausible transition kernels. We show how to solve the resulting program efficiently using sum-of-squares optimization and a Gaussian process envelope. Our approach lifts the need for restrictive assumptions on the system dynamics and uncertainty, and suggests an improvement in the sample complexity of verifying the safety of a system on a tested case study compared to a state-of-the-art approach., Comment: 7 pages, 2 figures, accepted to American Control Conference (ACC) 2024

Recently, several direct Data-Driven Predictive Control (DDPC) methods have been proposed, advocating the possibility of designing predictive controllers from historical input-output trajectories without the need to identify a model. In this work, we show that these approaches are equivalent to an indirect approach. Reformulating the direct methods in terms of estimated parameters and covariance matrices allows us to give new insights into how they work in comparison with, for example, Subspace Predictive Control (SPC). In particular, we show that for unconstrained problems the direct methods are equivalent to SPC with a reduced weight on the tracking cost. Via a numerical experiment, motivated by the reformulation, we also illustrate why the performance of direct DDPC methods with fixed regularization tends to degrade as the number of training samples increases., Comment: \c{opyright} 2024 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works

In this paper, we propose an extension for semi-supervised Minimum Sum-of-Squares Clustering (MSSC) problems of MDEClust, a memetic framework based on the Differential Evolution paradigm for unsupervised clustering. In semi-supervised MSSC, background knowledge is available in the form of (instance-level) "must-link" and "cannot-link" constraints, each of which indicating if two dataset points should be associated to the same or to a different cluster, respectively. The presence of such constraints makes the problem at least as hard as its unsupervised version and, as a consequence, some framework operations need to be carefully designed to handle this additional complexity: for instance, it is no more true that each point is associated to its nearest cluster center. As far as we know, our new framework, called S-MDEClust, represents the first memetic methodology designed to generate a (hopefully) optimal feasible solution for semi-supervised MSSC problems. Results of thorough computational experiments on a set of well-known as well as synthetic datasets show the effectiveness and efficiency of our proposal.

Atom probe tomography is frequently employed to characterize the elemental distribution in solids with atomic resolution. Here we review and discuss the potential of this technique to locally probe chemical bonds. Two processes characterize the bond rupture in laser-assisted field emission, the probability of molecular ions, i.e. the probability that molecular ions (PMI) are evaporated instead of single (atomic) ions, and the probability of multiple events, i.e. the correlated field-evaporation of more than a single fragment (PME) upon laser- or voltage pulse excitation. Here we demonstrate that one can clearly distinguish solids with metallic, covalent, and metavalent bonds based on their bond rupture, i.e. their PME and PMI values. Differences in the field penetration depth can largely explain these differences in bond breaking. These findings open new avenues in understanding and designing advanced materials, since they allow a quantification of bonds in solids on a nanometer scale, as will be shown for several examples. These possibilities would even justify calling the present approach bonding probe tomography (BPT).

This paper presents advanced techniques of training diffusion policies for offline reinforcement learning (RL). At the core is a mean-reverting stochastic differential equation (SDE) that transfers a complex action distribution into a standard Gaussian and then samples actions conditioned on the environment state with a corresponding reverse-time SDE, like a typical diffusion policy. We show that such an SDE has a solution that we can use to calculate the log probability of the policy, yielding an entropy regularizer that improves the exploration of offline datasets. To mitigate the impact of inaccurate value functions from out-of-distribution data points, we further propose to learn the lower confidence bound of Q-ensembles for more robust policy improvement. By combining the entropy-regularized diffusion policy with Q-ensembles in offline RL, our method achieves state-of-the-art performance on most tasks in D4RL benchmarks. Code is available at https://github.com/ruoqizzz/Entropy-Regularized-Diffusion-Policy-with-QEnsemble.

In this paper, we deal with batch Bayesian Optimization (Bayes-Opt) problems over a box and we propose a novel bi-objective optimization (BOO) acquisition strategy to sample points where to evaluate the objective function. The BOO problem involves the Gaussian Process posterior mean and variance functions, which, in most of the acquisition strategies from the literature, are generally used in combination, frequently through scalarization. However, such scalarization could compromise the Bayes-Opt process performance, as getting the desired trade-off between exploration and exploitation is not trivial in most cases. We instead aim to reconstruct the Pareto front of the BOO problem based on optimizing both the posterior mean as well as the variance, thus generating multiple trade-offs without any a priori knowledge. The reconstruction is performed through the Non-dominated Sorting Memetic Algorithm (NSMA), recently proposed in the literature and proved to be effective in solving hard MOO problems. Finally, we present two clustering approaches, each of them operating on a different space, to select potentially optimal points from the Pareto front. We compare our methodology with well-known acquisition strategies from the literature, showing its effectiveness on a wide set of experiments.

Electrodermal activity (EDA) reflects changes in skin conductance, which are closely tied to human psychophysiological states. For example, EDA sensors can assess stress, cognitive workload, arousal, or other measures tied to the sympathetic nervous system for interactive human-centered applications. Yet, current limitations involve the complex attachment and proper skin contact with EDA sensors. This paper explores the concept of 3D printing electrodes for EDA measurements, integrating sensors into arbitrary 3D-printed objects, alleviating the need for complex assembly and attachment. We examine the adaptation of conventional EDA circuits for 3D-printed electrodes, assessing different electrode shapes and their impact on the sensing accuracy. A user study (N=6) revealed that 3D-printed electrodes can measure EDA with similar accuracy, suggesting larger contact areas for improved precision. We derive design implications to facilitate the integration of EDA sensors into 3D-printed devices to foster diverse integration into everyday objects for prototyping physiological interfaces.

Accurate and comprehensive semantic segmentation of Bird's Eye View (BEV) is essential for ensuring safe and proactive navigation in autonomous driving. Although cooperative perception has exceeded the detection capabilities of single-agent systems, prevalent camera-based algorithms in cooperative perception neglect valuable information derived from historical observations. This limitation becomes critical during sensor failures or communication issues as cooperative perception reverts to single-agent perception, leading to degraded performance and incomplete BEV segmentation maps. This paper introduces TempCoBEV, a temporal module designed to incorporate historical cues into current observations, thereby improving the quality and reliability of BEV map segmentations. We propose an importance-guided attention architecture to effectively integrate temporal information that prioritizes relevant properties for BEV map segmentation. TempCoBEV is an independent temporal module that seamlessly integrates into state-of-the-art camera-based cooperative perception models. We demonstrate through extensive experiments on the OPV2V dataset that TempCoBEV performs better than non-temporal models in predicting current and future BEV map segmentations, particularly in scenarios involving communication failures. We show the efficacy of TempCoBEV and its capability to integrate historical cues into the current BEV map, improving predictions under optimal communication conditions by up to 2% and under communication failures by up to 19%. The code will be published on GitHub.

We calculate, via variational techniques, single- and two-photon Rydberg microwave transitions, as well as scalar and tensor polarizabilities of sodium atom using the parametric one-electron valence potential, including the spin-orbit coupling. The trial function is expanded in a basis set of optimized Slater-type orbitals, resulting in highly accurate and converged eigen-energies up to $n=60$. We focus our studies on the microwave band 90-150 GHz, due to its relevance to laser excitation in the Earth's upper-atmospheric sodium layer for wavelength-dependent radiometry and polarimetry, as precise microwave polarimetry in this band is an important source of systematic uncertainty in searches for signatures of primordial gravitational waves within the anisotropic polarization pattern of photons from the cosmic microwave background. We present the most efficient transition coefficients in this range, as well as the scalar and tensor polarizabilities compared with available experimental and theoretical data.

When deploying machine learning algorithms in the real world, guaranteeing safety is an essential asset. Existing safe learning approaches typically consider continuous variables, i.e., regression tasks. However, in practice, robotic systems are also subject to discrete, external environmental changes, e.g., having to carry objects of certain weights or operating on frozen, wet, or dry surfaces. Such influences can be modeled as discrete context variables. In the existing literature, such contexts are, if considered, mostly assumed to be known. In this work, we drop this assumption and show how we can perform safe learning when we cannot directly measure the context variables. To achieve this, we derive frequentist guarantees for multi-class classification, allowing us to estimate the current context from measurements. Further, we propose an approach for identifying contexts through experiments. We discuss under which conditions we can retain theoretical guarantees and demonstrate the applicability of our algorithm on a Furuta pendulum with camera measurements of different weights that serve as contexts., Comment: Accepted final version to appear in the IEEE Transactions on Robotics

Robots are ubiquitous in small-to-large-scale manufacturers. While collaborative robots (cobots) have significant potential in these settings due to their flexibility and ease of use, proper integration is critical to realize their full potential. Specifically, cobots need to be integrated in ways that utilize their strengths, improve manufacturing performance, and facilitate use in concert with human workers. Effective integration requires careful consideration and the knowledge of roboticists, manufacturing engineers, and business administrators. We propose an approach involving the stages of planning, analysis, development, and presentation, to inform manufacturers about cobot integration within their facilities prior to the integration process. We contextualize our approach in a case study with an SME collaborator and discuss insights learned., Comment: 9 pages, 9 figures. To be published in Proceedings of the 2024 ACM/IEEE International Conference on Human-Robot Interaction (HRI '24)