Search

This paper addresses the critical issue of psychological safety in the design and operation of autonomous vehicles, which are increasingly integrated with artificial intelligence technologies. While traditional safety standards focus primarily on physical safety, this paper emphasizes the psychological implications that arise from human interactions with autonomous vehicles, highlighting the importance of trust and perceived risk as significant factors influencing user acceptance. Through a review of existing safety techniques, the paper defines psychological safety in the context of autonomous vehicles, proposes a risk model to identify and assess psychological risks, and adopts a system-theoretic analysis method. The paper illustrates the potential psychological hazards using a scenario involving a family's experience with an autonomous vehicle, aiming to systematically evaluate situations that could lead to psychological harm. By establishing a framework that incorporates psychological safety alongside physical safety, the paper contributes to the broader discourse on the safe deployment of autonomous vehicle and aims to guide future developments in user-cantered design and regulatory practices.

Monolithic large language models (LLMs) like GPT-4 have paved the way for modern generative AI applications. Training, serving, and maintaining monolithic LLMs at scale, however, remains prohibitively expensive and challenging. The disproportionate increase in compute-to-memory ratio of modern AI accelerators have created a memory wall, necessitating new methods to deploy AI. Composition of Experts (CoE) is an alternative modular approach that lowers the cost and complexity of training and serving. However, this approach presents two key challenges when using conventional hardware: (1) without fused operations, smaller models have lower operational intensity, which makes high utilization more challenging to achieve; and (2) hosting a large number of models can be either prohibitively expensive or slow when dynamically switching between them. In this paper, we describe how combining CoE, streaming dataflow, and a three-tier memory system scales the AI memory wall. We describe Samba-CoE, a CoE system with 150 experts and a trillion total parameters. We deploy Samba-CoE on the SambaNova SN40L Reconfigurable Dataflow Unit (RDU) - a commercial dataflow accelerator architecture that has been co-designed for enterprise inference and training applications. The chip introduces a new three-tier memory system with on-chip distributed SRAM, on-package HBM, and off-package DDR DRAM. A dedicated inter-RDU network enables scaling up and out over multiple sockets. We demonstrate speedups ranging from 2$\times$ to 13$\times$ on various benchmarks running on eight RDU sockets compared with an unfused baseline. We show that for CoE inference deployments, the 8-socket RDU Node reduces machine footprint by up to 19$\times$, speeds up model switching time by 15$\times$ to 31$\times$, and achieves an overall speedup of 3.7$\times$ over a DGX H100 and 6.6$\times$ over a DGX A100., Comment: 2024 57th IEEE/ACM International Symposium on Microarchitecture (MICRO)

We present the results from the first two years of the Planet Hunters NGTS citizen science project, which searches for transiting planet candidates in data from the Next Generation Transit Survey (NGTS) by enlisting the help of members of the general public. Over 8,000 registered volunteers reviewed 138,198 light curves from the NGTS Public Data Releases 1 and 2. We utilize a user weighting scheme to combine the classifications of multiple users to identify the most promising planet candidates not initially discovered by the NGTS team. We highlight the five most interesting planet candidates detected through this search, which are all candidate short-period giant planets. This includes the TIC-165227846 system that, if confirmed, would be the lowest-mass star to host a close-in giant planet. We assess the detection efficiency of the project by determining the number of confirmed planets from the NASA Exoplanet Archive and TESS Objects of Interest (TOIs) successfully recovered by this search and find that 74% of confirmed planets and 63% of TOIs detected by NGTS are recovered by the Planet Hunters NGTS project. The identification of new planet candidates shows that the citizen science approach can provide a complementary method to the detection of exoplanets with ground-based surveys such as NGTS., Comment: 42 pages, 20 figures, 17 tables. To be published in AJ

To date, thousands of planets have been discovered, but there are regions of the orbital parameter space that are still bare. An example is the short period and intermediate mass/radius space known as the Neptunian desert, where planets should be easy to find but discoveries remain few. This suggests unusual formation and evolution processes are responsible for the planets residing here. We present the discovery of TOI-332 b, a planet with an ultra-short period of $0.78$ d that sits firmly within the desert. It orbits a K0 dwarf with an effective temperature of $5251 \pm 71$ K. TOI-332 b has a radius of $3.20^{+0.16}_{-0.12}$ R$_{\oplus}$, smaller than that of Neptune, but an unusually large mass of $57.2 \pm 1.6$ M$_{\oplus}$. It has one of the highest densities of any Neptune-sized planet discovered thus far at $9.6^{+1.1}_{-1.3}$ gcm$^{-3}$. A 4-layer internal structure model indicates it likely has a negligible hydrogen-helium envelope, something only found for a small handful of planets this massive, and so TOI-332 b presents an interesting challenge to planetary formation theories. We find that photoevaporation cannot account for the mass loss required to strip this planet of the Jupiter-like envelope it would have been expected to accrete. We need to look towards other scenarios, such as high-eccentricity migration, giant impacts, or gap opening in the protoplanetary disc, to try and explain this unusual discovery., Comment: 20 pages, 14 figures, 6 tables, accepted for publication in MNRAS

The generalisation of proper time, as an alternative to models with extra dimensions of space, has been proposed as the source of the elementary structures of matter. Direct connections with the Standard Model of particle physics together with dark sector candidates can be attained in this manner, through the additional components and symmetry breaking pattern identified on extracting the substructure of the local 4-dimensional spacetime form. In this paper we focus more on the extended and global properties of the theory, both in relation to amalgamating gravity with quantum theory and to broader cosmological issues. Contrast will be made with both early S-matrix bootstrap models and more recent cosmological bootstrap techniques. With time playing a central role in the new theory we describe how a 'universal bootstrap' can be constructed and elaborate upon the implications., Comment: 84 pages, 10 figures

Although unsupervised domain adaptation methods have achieved remarkable performance in semantic scene segmentation in visual perception for self-driving cars, these approaches remain impractical in real-world use cases. In practice, the segmentation models may encounter new data that have not been seen yet. Also, the previous data training of segmentation models may be inaccessible due to privacy problems. Therefore, to address these problems, in this work, we propose a Continual Unsupervised Domain Adaptation (CONDA) approach that allows the model to continuously learn and adapt with respect to the presence of the new data. Moreover, our proposed approach is designed without the requirement of accessing previous training data. To avoid the catastrophic forgetting problem and maintain the performance of the segmentation models, we present a novel Bijective Maximum Likelihood loss to impose the constraint of predicted segmentation distribution shifts. The experimental results on the benchmark of continual unsupervised domain adaptation have shown the advanced performance of the proposed CONDA method., Comment: Accepted to CVPRW 2024

We report the discovery of three new hot Jupiters with the Next Generation Transit Survey (NGTS) as well as updated parameters for HATS-54b, which was independently discovered by NGTS. NGTS-23b, NGTS-24b and NGTS-25b have orbital periods of 4.076, 3.468, and 2.823 days and orbit G-, F- and K-type stars, respectively. NGTS-24 and HATS-54 appear close to transitioning off the main-sequence (if they are not already doing so), and therefore are interesting targets given the observed lack of Hot Jupiters around sub-giant stars. By considering the host star luminosities and the planets' small orbital separations (0.037 - 0.050 au), we find that all four hot Jupiters are above the minimum irradiance threshold for inflation mechanisms to be effective. NGTS-23b has a mass of 0.61 $M_{J}$ and radius of 1.27 $R_{J}$ and is likely inflated. With a radius of 1.21 $R_{J}$ and mass of 0.52 $M_{J}$, NGTS-24b has a radius larger than expected from non-inflated models but its radius is smaller than the predicted radius from current Bayesian inflationary models. Finally, NGTS-25b is intermediate between the inflated and non-inflated cases, having a mass of 0.64 $M_{J}$ and a radius of 1.02 $R_{J}$. The physical processes driving radius inflation remain poorly understood, and by building the sample of hot Jupiters we can aim to identify the additional controlling parameters, such as metallicity and stellar age., Comment: 17 pages, 8 figures. Accepted for publication in MNRAS

A central aim of theoretical physics is to account for the structure of matter at the most elementary level as underlying the Standard Model of particle physics, and ideally also as a basis for a substantial dark sector, as distributed in 4-dimensional spacetime. A broad class of theories augment the global 4-dimensional spacetime arena itself to a higher-dimensional spacetime structure, with the properties of matter then deriving from the properties of the extra spatial dimensions. We motivate an alternative approach in which we begin with the local structure of 4-dimensional spacetime and augment the local form for a proper time interval. The symmetries and properties of the residual components, over the local 4-dimensional spacetime form, are found to exhibit appropriate features to account for the visible Standard Model sector while in parallel direct connections can also be identified with dark sector models., Comment: 27 pages, 3 figures

When the first CubeSats were launched nearly two decades ago, few people believed that the miniature satellites would likely prove to be a useful scientific tool. Skeptics abounded. However, the last decade has seen the highly successful implementation of space missions that make creative and innovative use of fast-advancing CubeSat and small satellite technology to carry out important science experiments and missions. Several projects now have used CubeSats to obtain first-of-their-kind observations and findings that have formed the basis for high-profile engineering and science publications, thereby establishing without doubt the scientific value and broad utility of CubeSats. In this paper, we describe recent achievements and lessons learned from a representative selection of successful CubeSat missions with a space weather focus. We conclude that these missions were successful in part because their limited resources promoted not only mission focus but also appropriate risk-taking for comparatively high science return. Quantitative analysis of refereed publications from these CubeSat missions and several larger missions reveals that mission outcome metrics compare favorably when publication number is normalized by mission cost or if expressed as a weighted net scientific impact of all mission publications., Comment: 35 pages, 5 figures, 1 table; accepted for publication by Space Weather Journal