Filetype
Bayesian optimization is a widely used method for optimizing expensive black-box functions, with Expected Improvement being one of the most commonly used acquisition functions. In contrast, information-theoretic acquisition functions aim to reduce uncertainty about the function's optimum and are often considered fundamentally distinct from EI. In this work, we challenge this prevailing perspective
Recent work reported that simple Bayesian optimization methods perform well for high-dimensional real-world tasks, seemingly contradicting prior work and tribal knowledge. This paper investigates the 'why'. We identify fundamental challenges that arise in high-dimensional Bayesian optimization and explain why recent methods succeed. Our analysis shows that vanishing gradients caused by Gaussian pr
Recent technologies of district heating and cooling (DHC) systems integrate renewable energy sources and connect multiple heat producers and consumers to the same network. The resulting interactions may therefore cause the fluid in the pipe, or the energy carried by the system, to flow in both directions. This article presents a newly developed analytical method for evaluating the hydraulic states
We recently proposed a quantitative group testing (GT) scheme with low-complexity peeling decoding based on low-density parity-check (LDPC) codes. Based on finite length simulations and a density evolution analysis we were able to demonstrate that simple (dv,dc)-regular LDPC codes can be more efficient for GT than existing generalized LDPC (GLDPC) code constructions based on BCH component codes. E
Future wireless systems are envisioned to be able to deliver ultra-reliable and low-latency communication.The third generation partner project (3GPP) has identified three different usage scenarios such as enhanced mobile broadband, massive machine-type communication, and ultra-reliable low latency communication.In each of those scenarios, several services and applications can be implemented, e.g.
Delivery care in Sweden is very safe, the incidence of maternal death is 5/100 000 and perinatal death <2/100 000. The perinatal death rate among babies born at or after week 41+0 decreased from 0.17 to 0.09% (p<0.001) and obstetric anal sphincter injuries have decreased from 3.5% to 2.6%; however interventions such as induction of labour increase. A national project to improve safety for newborns
The IEEE 802.11 standard, often referred to as Wi-Fi, underpins wireless networking applications around the world that impact our daily lives, such as wireless access to the Internet from offices, homes, airports, hotels, restaurants, trains, and aircraft. Today's laptops, tablets, and smartphones are typically equipped with at least one IEEE 802.11 radio. IEEE 802.11 standards have enabled a whol
We show that spatially coupled low-density parity-check (LDPC) codes yield robust performance over changing intersymbol interfere (ISI) channels with optimal and suboptimal detectors. We compare the performance with classical LDPC code design which involves optimizing the degree distribution for a given (known) channel. We demonstrate that these classical schemes, despite working very good when de
Formal models for the safety validation of autonomous vehicles have become increasingly important. To this end, we present a safety framework for longitudinal automated driving. This framework allows calculating minimum safe inter-vehicular distances for arbitrary ego vehicle control policies. We use this framework to enhance the Responsibility-Sensitive Safety (RSS) model and models based on it,
This issue marks the start of my term as department editor for the “Requirements” column. I very much look forward to exploring contemporary aspects of requirements and requirements engineering (RE) in the coming years! As an institute researcher with RISE, I primarily work in strictly regulated domains, in which requirements are cornerstones in the development activities. Please check my introduc
Memories may integrate elements experienced in different events. For instance, meeting a woman leaving her house, and later meeting another woman entering the same house, may allow us to infer that the two women live together. Such memory representations are thought to rely on integrative encoding mechanisms, allowing us to make inferences about the world and generalize knowledge to entirely new s
ObjectivesThe formation of episodic memories is critically determined by how we visually sample the world over time via sequences of eye movements. Nonetheless, in the neuroscience of human memory, memory encoding has almost exclusively been studied in experimental paradigms where the study material is presented in a single fixed location on the screen, and where eye movements are treated as artif
Episodic memory enables mental time travel, allowing us to relive specific, personally experienced events tied in time and place. This feat of human memory is considered to be dependent on the reinstatement of the cortical patterns that were active at the time of encoding. A growing body of recent literature has provided support for this idea by showing that retrieval success co-varies with the ne
In this paper, we examine variable node (VN) doping to mitigate the error propagation problem in sliding window decoding (SWD) of spatially coupled LDPC (SC-LDPC) codes from the point of view of the encoding process. More specifically, in order to simplify the process of generating an encoded sequence with some number of doped code bits, we propose to employ systematic encoding and to limit doping
The commercial availability of low-cost millimeterwave (mmWave) communication and radar devices is starting to improve the adoption of such technologies in consumer markets, paving the way for large-scale and dense deployments in fifthgeneration (5G)-and-beyond as well as 6G networks. At the same time, pervasive mmWave access will enable device localization and device-free sensing with unprecedent
Cell-free massive multiple-input multiple-output (MIMO) consists of a large set of distributed access points (APs) serving a number of users. The APs can be far from each other, and they can also have a big number of antennas. Thus, decentralized architectures have to be considered so as to reduce the interconnection bandwidth to a central processing unit (CPU) and make the system scalable. On the
SMIRK is a pedestrian automatic emergency braking system that facilitates research on safety-critical systems embedding machine learning components. As a fully transparent driver-assistance system, SMIRK can support future research on trustworthy AI systems, e.g., verification & validation, requirements engineering, and testing. SMIRK is implemented for the simulator ESI Pro-SiVIC with core co
Robot skills systems are meant to reduce robot setup time for new manufacturing tasks. Yet, for dexterous, contact-rich tasks, it is often difficult to find the right skill parameters. One strategy is to learn these parameters by allowing the robot system to learn directly on the task. For a learning problem, a robot operator can typically specify the type and range of values of the parameters. Ne
We propose a generalisation of a behaviour tree and motiongenerator based robot arm policy representation for learning and solving tasks such as contact-rich tasks like peg insertion or pushing an object. We use planning to generate skill sequences needed to execute these tasks and rely on reinforcement learning to obtain parameters of the policy. We assume gaussian processes as a suitable method
We propose a pipeline to setup up robot systems for new tasks using symbolic planning, reasoning and reinforcement learning. Furthermore, we
