Search

In many practical applications, large language models (LLMs) need to incorporate new knowledge not present in their pre-training data. The primary methods for this are fine-tuning and retrieval-augmented generation (RAG). Although RAG has emerged as the industry standard for knowledge injection, fine-tuning has not yet achieved comparable success. In this paper, we propose a new fine-tuning technique for learning new knowledge and show that it can reach the performance of RAG. The proposed method is based on the self-distillation approach, which we call prompt distillation. First, we generate question-answer pairs about the new knowledge. Then, we fine-tune a student model on the question-answer pairs to imitate the output distributions of a teacher model, which additionally receives the new knowledge in its prompt. The student model is identical to the teacher, except it is equipped with a LoRA adapter. This training procedure facilitates distilling the new knowledge from the teacher's prompt into the student's weights., Comment: Preprint

We find identities involving differential operators in the generic artinian reduction of the Stanley-Reisner ring of a simplicial sphere in any positive characteristic. These identities generalize the characteristic 2 identities used by Papadakis and Petrotou to give a proof of the algebraic g-conjecture. We use these identities to prove the anisotropy of certain forms on the generic artinian reduction of the Stanley-Reisner ring and to prove weak Lefschetz results.

Time delay estimation or Time-Difference-Of-Arrival estimates is a critical component for multiple localization applications such as multilateration, direction of arrival, and self-calibration. The task is to estimate the time difference between a signal arriving at two different sensors. For the audio sensor modality, most current systems are based on classical methods such as the Generalized Cross-Correlation Phase Transform (GCC-PHAT) method. In this paper we demonstrate that learning based methods can, even based on synthetic data, significantly outperform GCC-PHAT on novel real world data. To overcome the lack of data with ground truth for the task, we train our model on a simulated dataset which is sufficiently large and varied, and that captures the relevant characteristics of the real world problem. We provide our trained model, SONNET (Simulation Optimized Neural Network Estimator of Timeshifts), which is runnable in real-time and works on novel data out of the box for many real data applications, i.e. without re-training. We further demonstrate greatly improved performance on the downstream task of self-calibration when using our model compared to classical methods.

Participatory citizen platforms are innovative solutions to digitally better engage citizens in policy-making and deliberative democracy in general. Although these platforms have been used also in an engineering context, thus far, there is no existing work for connecting the platforms to requirements engineering. The present paper fills this notable gap. In addition to discussing the platforms in conjunction with requirements engineering, the paper elaborates potential advantages and disadvantages, thus paving the way for a future pilot study in a software engineering context. With these engineering tenets, the paper also contributes to the research of large socio-technical software systems in a public sector context, including their implementation and governance., Comment: Submitted to REFSQ

The statistics of transmitted photons in microwave cavities play a foundational role in microwave quantum optics and its technological applications. By utilizing quantum mechanical phase-space methods, we here develop a general theory of the photon counting statistics in Gaussian bosonic networks consisting of driven cavities with beamsplitter interactions and two-mode-squeezing. The dynamics of the network can be captured by a Lyapunov equation for the covariance matrix of the cavity fields, which generalizes to a Riccati equation, when counting fields are included. By solving the Riccati equation, we obtain the statistics of emitted and absorbed photons as well as the time-dependent correlations encoded in waiting time distributions and second-order coherence functions. To illustrate our theoretical framework, we first apply it to a simple linear network consisting of two coupled cavities, for which we evaluate the photon cross-correlations and discuss connections between the photon emission statistics and the entanglement between the cavities. We then consider a bosonic circulator consisting of three coupled cavities, for which we investigate how a synthetic flux may affect the direction of the photon flow, similarly to recent experiments. Our general framework paves the way for systematic investigations of the photon counting statistics in Gaussian bosonic networks., Comment: 35 pages, 7 figures

This work presents a fully integrated tree-based combined exploration-planning algorithm: Exploration-RRT (ERRT). The algorithm is focused on providing real-time solutions for local exploration in a fully unknown and unstructured environment while directly incorporating exploratory behavior, robot-safe path planning, and robot actuation into the central problem. ERRT provides a complete sampling and tree-based solution for evaluating "where to go next" by considering a trade-off between maximizing information gain, and minimizing the distances travelled and the robot actuation along the path. The complete scheme is evaluated in extensive simulations, comparisons, as well as real-world field experiments in constrained and narrow subterranean and GPS-denied environments. The framework is fully ROS-integrated, straight-forward to use, and we open-source it at https://github.com/LTU-RAI/ExplorationRRT., Comment: 19 pages, 29 figures Transactions on Robotics

This paper introduces the notion of a universal plan, which when executed, is guaranteed to solve all planning problems in a category, regardless of the obstacles, initial state, and goal set. Such plans are specified as a deterministic sequence of actions that are blindly applied without any sensor feedback. Thus, they can be considered as pure exploration in a reinforcement learning context, and we show that with basic memory requirements, they even yield optimal plans. Building upon results in number theory and theory of automata, we provide universal plans both for discrete and continuous (motion) planning and prove their (semi)completeness. The concepts are applied and illustrated through simulation studies, and several directions for future research are sketched., Comment: 20 pages, 4 figures

It is desirable that a given continued fraction algorithm is simple in the sense that the possible representations can be characterized in an easy way. In this context the so-called finite range condition plays a prominent role. We show that this condition holds for complex $\boldsymbol{\alpha}$-Hurwitz algorithms with parameters $\boldsymbol{\alpha}\in\mathbb{Q}^2$. This is equivalent to the existence of certain finite partitions related to these algorithms and lies at the root of explorations into their Diophantine properties. Our result provides a partial answer to a recent question formulated by Lukyanenko and Vandehey., Comment: 12 pages, 6 figures

A fundamental feature of quantum mechanics is that there are observables which can be measured jointly only when some noise is added to them. Their sharp versions are said to be incompatible. In this work we investigate time-continuous joint qubit observables induced by a indirect time-continuous measurements. In particular we study a paradigmatic situation where a qubit is interacting with a mode of light in a cavity and the light escaping the cavity is continuously monitored. We find that the properties of the qubit observables can be tuned by changing the type of the monitoring scheme or by tuning the initial state of the cavity. We observe that homodyning two orthogonal quadratures produces an optimal pair of biased jointly measurable qubit observables., Comment: 11 pages, 6 figures. Working paper. Comments are very welcome!

When designing a robot's internal system, one often makes assumptions about the structure of the intended environment of the robot. One may even assign meaning to various internal components of the robot in terms of expected environmental correlates. In this paper we want to make the distinction between robot's internal and external worlds clear-cut. Can the robot learn about its environment, relying only on internally available information, including the sensor data? Are there mathematical conditions on the internal robot system which can be internally verified and make the robot's internal system mirror the structure of the environment? We prove that sufficiency is such a mathematical principle, and mathematically describe the emergence of the robot's internal structure isomorphic or bisimulation equivalent to that of the environment. A connection to the free-energy principle is established, when sufficiency is interpreted as a limit case of surprise minimization. As such, we show that surprise minimization leads to having an internal model isomorphic to the environment. This also parallels the Good Regulator Principle which states that controlling a system sufficiently well means having a model of it. Unlike the mentioned theories, ours is discrete, and non-probabilistic., Comment: 19 pages, 2 figures

Attention-based transformers have become the standard architecture in many deep learning fields, primarily due to their ability to model long-range dependencies and handle variable-length input sequences. However, the attention mechanism with its quadratic complexity is a significant bottleneck in the transformer architecture. This algorithm is only uni-directional in the decoder and converges to a static pattern in over-parametrized decoder-only models. I address this issue by developing a generative function as attention or activation replacement. It still has the auto-regressive character by comparing each token with the previous one. In my test setting with nanoGPT this yields a smaller loss while having a smaller model. The loss further drops by incorporating an average context vector. This concept of attention replacement is distributed under the GNU AGPL v3 license at https://gitlab.com/Bachstelze/causal_generation., Comment: 6 pages, 4 figures

Fairness is one of the most commonly identified ethical principles in existing AI guidelines, and the development of fair AI-enabled systems is required by new and emerging AI regulation. But most approaches to addressing the fairness of AI-enabled systems are limited in scope in two significant ways: their substantive content focuses on statistical measures of fairness, and they do not emphasize the need to identify and address fairness considerations across the whole AI lifecycle. Our contribution is to present an assurance framework and tool that can enable a practical and transparent method for widening the scope of fairness considerations across the AI lifecycle and move the discussion beyond mere statistical notions of fairness to consider a richer analysis in a practical and context-dependent manner. To illustrate this approach, we first describe and then apply the framework of Trustworthy and Ethical Assurance (TEA) to an AI-enabled clinical diagnostic support system (CDSS) whose purpose is to help clinicians predict the risk of developing hypertension in patients with Type 2 diabetes, a context in which several fairness considerations arise (e.g., discrimination against patient subgroups). This is supplemented by an open-source tool and a fairness considerations map to help facilitate reasoning about the fairness of AI-enabled systems in a participatory way. In short, by using a shared framework for identifying, documenting and justifying fairness considerations, and then using this deliberative exercise to structure an assurance case, research on AI fairness becomes reusable and generalizable for others in the ethical AI community and for sharing best practices for achieving fairness and equity in digital health and healthcare in particular., Comment: 18 pages, 2 figures

This letter explores an implementation of a novel thin film 1-by-4 reconfigurable intelligent surface (RIS) designed for future communication and sensing scenarios. Utilizing cost-effective inkjet printing methods and additive manufacturing, our approach significantly simplifies the RIS construction process and reduces production costs. The RIS, fabricated on a flexible and lightweight polyethylene terephthalate (PET) substrate, integrates antennas, switching circuitry, and a microcontroller unit (MCU), without a ground shield. This setup enables individual and simultaneous control of each RIS element, manipulating the captured carrier signal by reflecting and refracting its dominant harmonics. Beams of the harmonics can be steered to multiple desired directions at both front and back sides of the surface. Measurement results of the beam steering show that the RIS has the potential to enable RIS-aided communication and sensing applications., Comment: 5 pages, 4 figures, IEEE letter

Learning representations for reinforcement learning (RL) has shown much promise for continuous control. We propose an efficient representation learning method using only a self-supervised latent-state consistency loss. Our approach employs an encoder and a dynamics model to map observations to latent states and predict future latent states, respectively. We achieve high performance and prevent representation collapse by quantizing the latent representation such that the rank of the representation is empirically preserved. Our method, named iQRL: implicitly Quantized Reinforcement Learning, is straightforward, compatible with any model-free RL algorithm, and demonstrates excellent performance by outperforming other recently proposed representation learning methods in continuous control benchmarks from DeepMind Control Suite., Comment: 9 pages, 11 figures

Data breaches refer to unauthorized accesses to data. Typically but not always, data breaches are about cyber crime. An organization facing such a crime is often also in a crisis situation. Therefore, organizations should prepare also for data breaches in their crisis management procedures. These procedures should include also crisis communication plans. To this end, this paper examines data breach crisis communication strategies and their practical executions. The background comes from the vibrant crisis communication research domain. According to a few qualitative case studies from Finland, the conventional wisdom holds well; the successful cases indicate communicating early, taking responsibility, offering an apology, and notifying public authorities. The unsuccessful cases show varying degrees of the reverse, including shifting of blame, positioning of an organization as a victim, and failing to notify public authorities. With these qualitative insights, the paper contributes to the research domain by focusing specifically on data breach crises, their peculiarities, and their management, including with respect to European regulations that have been neglected in existing crisis communication research., Comment: Submitted

This paper explores the novel topic of data breach journalism and data breach news through the case of databreaches.net, a news outlet dedicated to data breaches and related cyber crime. Motivated by the issues in traditional crime news and crime journalism, the case is explored by the means of text mining. According to the results, the outlet has kept a steady publishing pace, mainly focusing on plain and short reporting but with generally high-quality source material for the news articles. Despite these characteristics, the news articles exhibit fairly strong sentiments, which is partially expected due to the presence of emotionally laden crime and the long history of sensationalism in crime news. The news site has also covered the full scope of data breaches, although many of these are fairly traditional, exposing personal identifiers and financial details of the victims. Also hospitals and the healthcare sector stand out. With these results, the paper advances the study of data breaches by considering these from the perspective of media and journalism., Comment: Proceedings of the 19th International Conference on Availability, Reliability and Security (ARES 2024), Vienna, ACM, pp. 1-9

To connect conformal field theories (CFT) to probabilistic lattice models, recent works [HKV22, Ada23] have introduced a novel definition of local fields of the lattice models. Local fields in this picture are probabilistically concrete: they are built from random variables in the model. The key insight is that discrete complex analysis ideas allow to equip the space of local fields with the main structure of a CFT: a representation of the Virasoro algebra. In this article, for the first time, we fully analyze the structure of the space of local fields of a lattice model as a representation, and use this to establish a one-to-one correspondence between the local fields of a lattice model and those of a conformal field theory. The CFT we consider is probabilistically realized in terms of the gradient of the Gaussian Free Field (GFF). Its space of local fields is just a bosonic Fock space for two chiral symmetry algebras. The corresponding lattice model is the discrete Gaussian Free Field. Our first main result is that the space of local fields of polynomials in the gradient of the discrete GFF is isomorphic to the Fock space. These local fields make sense with both Dirichlet and Neumann boundary conditions. Our second main result is that with the appropriate renormalization, correlation functions of local fields of the discrete GFF converge in the scaling limit to the correlation functions of the CFT. The renormalization needed is, conceptually correctly, according to the eigenvalue of the Virasoro generator $L_0 + \bar{L}_0$ on the local field., Comment: 84 pages, 14 figures, comments welcome! (v2: minor edits and misprint corrections)

Our aim in this article is to compute the mixed volume of a matroid. We give two computations. The first one is based on the integration formula for complete fans given by Brion. The second computation is a step-by-step method using deletion of elements in the matroid as in Braden-Huh-Matherne-Proudfoot-Wang., Comment: 29 pages, 1 figure

For countably infinite IFSs on $\mathbb R^2$ consisting of affine contractions with diagonal linear parts, we give conditions under which the affinity dimension is an upper bound for the Hausdorff dimension and a lower bound for the lower box-counting dimension. Moreover, we identify a family of countably infinite IFSs for which the Hausdorff and affinity dimension are equal, and which have full dimension spectrum. The corresponding self-affine sets are related to restricted digit sets for signed L\"uroth expansions., Comment: 16 pages, 1 figure with 3 subfigures

We present a versatile NeRF-based simulator for testing autonomous driving (AD) software systems, designed with a focus on sensor-realistic closed-loop evaluation and the creation of safety-critical scenarios. The simulator learns from sequences of real-world driving sensor data and enables reconfigurations and renderings of new, unseen scenarios. In this work, we use our simulator to test the responses of AD models to safety-critical scenarios inspired by the European New Car Assessment Programme (Euro NCAP). Our evaluation reveals that, while state-of-the-art end-to-end planners excel in nominal driving scenarios in an open-loop setting, they exhibit critical flaws when navigating our safety-critical scenarios in a closed-loop setting. This highlights the need for advancements in the safety and real-world usability of end-to-end planners. By publicly releasing our simulator and scenarios as an easy-to-run evaluation suite, we invite the research community to explore, refine, and validate their AD models in controlled, yet highly configurable and challenging sensor-realistic environments. Code and instructions can be found at https://github.com/atonderski/neuro-ncap

This paper makes the case that a powerful new discipline, which we term perception engineering, is steadily emerging. It follows from a progression of ideas that involve creating illusions, from historical paintings and film, to video games and virtual reality in modern times. Rather than creating physical artifacts such as bridges, airplanes, or computers, perception engineers create illusory perceptual experiences. The scope is defined over any agent that interacts with the physical world, including both biological organisms (humans, animals) and engineered systems (robots, autonomous systems). The key idea is that an agent, called a producer, alters the environment with the intent to alter the perceptual experience of another agent, called a receiver. Most importantly, the paper introduces a precise mathematical formulation of this process, based on the von Neumann-Morgenstern notion of information, to help scope and define the discipline. It is then applied to the cases of engineered and biological agents with discussion of its implications on existing fields such as virtual reality, robotics, and even social media. Finally, open challenges and opportunities for involvement are identified., Comment: 30 pages, 5 figures

Let $L=(L_d)_{d \in \mathbb N}$ be any ordered probability sequence, i.e., satisfying $0 < L_{d+1} \le L_d$ for each $d \in \mathbb N$ and $\sum_{d \in \mathbb N} L_d =1$. We construct sequences $A = (a_i)_{i \in \mathbb N}$ on the countably infinite alphabet $\mathbb N$ in which each possible block of digits $\alpha_1, \ldots, \alpha_k \in \mathbb N$, $k \in \mathbb N$, occurs with frequency $\prod_{d=1}^k L_{\alpha_d}$. In other words, we construct $L$-normal sequences. These sequences can then be projected to normal numbers in various affine number systems, such as real numbers $x \in [0,1]$ that are normal in GLS number systems that correspond to the sequence $L$ or higher dimensional variants. In particular, this construction provides a family of numbers that have a normal L\"uroth expansion., Comment: 21 pages, 3 figures

The 3GPP has recently conducted a study on the Ambient Internet of Things (AIoT), with a particular emphasis on examining backscatter communications as one of the primary techniques under consideration. Previous investigations into Ambient Backscatter Communications (AmBC) within the long term evolution (LTE) downlink have shown that it is feasible to utilize the user equipment channel estimator as a receiver for demodulating frequency shift keyed (FSK) messages transmitted by the backscatter devices. In practical deployment scenarios, the backscattered link often experiences a low signal-to-noise ratio, leading to subpar bit error rate (BER) performance in the case of uncoded transmissions. In this paper, we propose the adoption of the same convolutional coding methodology for backscatter links that is already employed for LTE downlink control signals. This approach facilitates the reuse of identical demodulation functions at the modem for both control signals and backscattered AIoT messages. To assess the performance of the proposed scheme, we conducted experiments utilizing real LTE downlink signals generated by a mobile operator within an office environment. When compared to uncoded FSK, convolutional channel coding delivers a notable gain of approximately 6 dB at a BER of $10^{-3}$. Consequently, the AmBC system demonstrates a high level of reliability, achieving a BER of $10^{-3}$ at a Signal-to-Noise Ratio (SNR) of 5 dB.

Building simulation environments for developing and testing autonomous vehicles necessitates that the simulators accurately model the statistical realism of the real-world environment, including the interaction with other vehicles driven by human drivers. To address this requirement, an accurate human behavior model is essential to incorporate the diversity and consistency of human driving behavior. We propose a mathematical framework for designing a data-driven simulation model that simulates human driving behavior more realistically than the currently used physics-based simulation models. Experiments conducted using the NGSIM dataset validate our hypothesis regarding the necessity of considering the complexity, diversity, and consistency of human driving behavior when aiming to develop realistic simulators.

We address the challenges in estimating 3D human poses from multiple views under occlusion and with limited overlapping views. We approach multi-view, single-person 3D human pose reconstruction as a regression problem and propose a novel encoder-decoder Transformer architecture to estimate 3D poses from multi-view 2D pose sequences. The encoder refines 2D skeleton joints detected across different views and times, fusing multi-view and temporal information through global self-attention. We enhance the encoder by incorporating a geometry-biased attention mechanism, effectively leveraging geometric relationships between views. Additionally, we use detection scores provided by the 2D pose detector to further guide the encoder's attention based on the reliability of the 2D detections. The decoder subsequently regresses the 3D pose sequence from these refined tokens, using pre-defined queries for each joint. To enhance the generalization of our method to unseen scenes and improve resilience to missing joints, we implement strategies including scene centering, synthetic views, and token dropout. We conduct extensive experiments on three benchmark public datasets, Human3.6M, CMU Panoptic and Occlusion-Persons. Our results demonstrate the efficacy of our approach, particularly in occluded scenes and when few views are available, which are traditionally challenging scenarios for triangulation-based methods., Comment: Accepted: 18th IEEE International Conference on Automatic Face and Gesture Recognition (FG 2024)