King's College, Aberdeen

About: King's College, Aberdeen is a based out in . It is known for research contribution in the topics: Poison control & Sedimentary depositional environment. The organization has 712 authors who have published 918 publications receiving 25421 citations. The organization is also known as: King's College, Aberdeen & The University and King's College of Aberdeen.

Application of a brain-inspired deep imitation learning algorithm in autonomous driving

Abstract: Autonomous driving has attracted great attention from both academics and industries. To realise autonomous driving, Deep Imitation Learning (DIL) is treated as one of the most promising solutions, because it improves autonomous driving systems by automatically learning a complex mapping from human driving data, compared to manually designing the driving policy. However, existing DIL methods cannot generalise well across domains, that is, a network trained on the data of source domain gives rise to poor generalisation on the data of target domain. In the present study, we propose a novel brain-inspired deep imitation method that builds on the evidence from human brain functions, to improve the generalisation ability of DNN so that autonomous driving systems can perform well in various scenarios. Specifically, humans have a strong generalisation ability which is beneficial from the structural and functional asymmetry of the two sides of the brain. Here, we design dual Neural Circuit Policy (NCP) architectures in DNN based on the asymmetry of human neural networks. Experimental results demonstrate that our brain-inspired method outperforms existing methods regarding generalisation when dealing with unseen data. Our source codes and pretrained models are available at https://github.com/Intenzo21/Brain-Inspired-Deep-Imitation-Learning-for-Autonomous-Driving-Systems .

Sub-LFL flame kernels in formic acid vapour/air mixtures

Abstract: among fire experts, scientists and consultants. Contributions to The Forum will not be refereed in the conventional sense, but will be subject to review by the Journal’s Editorial Board relative to appropriateness, clarity, timeliness, and scope of interest. The Editorial Board will be the sole judge of those contributions to be published. Opinions expressed, however, are those of the authors and not of the Editors or Technomic Publishing Company, Incorporated.

Variants of some of the Brauer-Fowler Theorems

Abstract: Brauer and Fowler noted restrictions on the structure of a finite group G in terms of the order of the centralizer of an involution t in G. We consider variants of these themes. We first note that for an arbitrary finite group G of even order, we have |G| is less than the number of conjugacy classes of the Fitting subgroup times the order of the centralizer to the fourth power of any involution in G. This result does require the classification of the finite simple groups. The groups SL(2,q) with q even shows that the exponent 4 cannot be replaced by any exponent less than 3. We do not know at present whether the exponent 4 can be improved in general, though we note that the exponent 3 suffices for almost simple groups G. We are however able to prove that every finite group $G$ of even order contains an involution u such that [G:F(G)] is less than the cube of the order of the centralizer of u. There is a dichotomy in the proof of this fact, as it reduces to proving two residual cases: one in which G is almost simple (where the classification of the finite simple groups is needed) and one when G has a Sylow 2-subgroup of order 2. For the latter result, the classification of finite simple groups is not needed (though the Feit-Thompson odd order theorem is). We also prove a very general result on fixed point spaces of involutions in finite irreducible linear groups which does not make use of the classification of the finite simple groups, and some other results on the existence of non-central elements (not necessarily involutions) with large centralizers in general finite groups. We also show (without the classification of finite simple groups) that if t is an involution in G and p is a prime divisor of [G:F(G)], then p is at most 1 plus the order of the centralizer of t (and this is best possible).