搜索结果：should

Our community believes that new domain-specific languages should be as general as possible to increase their impact. However, I argue in this essay that we should stop claiming generality for new domain-specific languages. More general domain-specific languages induce more boilerplate code. Moreover, domain-specific languages are co-developed with their applications in practice, and tend to be specific for these applications. Thus, I argue we should stop claiming generality in favor of documenting how domain-specific language based software development is beneficial to the overall software development process. The acceptance criteria for scientific literature should make the same shift: accepting good domain-specific language engineering practice, instead of the next language to rule them all.

There are two strong arguments in favor of vector-like leptons and quarks: Flavor Democracy call for them, and E6 GUT predicts existence of iso-singlet quarks and iso-doublet leptons. Vector-like quarks (VLQ) are extensively searched by ATLAS and CMS collaborations, but this is not the case for vector-like leptons (VLL), while they have actually similar status from phenomenology viewpoint. In this study we argue that vector-like leptons should be included into the new physics search programs of energy-frontier colliders. We consider production of vector-like partners of the first SM family leptons at the HL-LHC, HE-LHC, FCC, ILC, CLIC, Muon Collider, as well as, at ep and μ-p colliders. As for decays of vector-like leptons, we present branching ratios formulas to different channels for the most general case. Since there are many different production and decay channels for charged and neutral vector-like leptons, relevant studies should be done systematically. We invite the High Energy Physics community (both experimenters and phenomenologists) to actively participate in research on this topic.

With the growing popularity of deep-learning based NLP models, comes a need for interpretable systems. But what is interpretability, and what constitutes a high-quality interpretation? In this opinion piece we reflect on the current state of interpretability evaluation research. We call for more clearly differentiating between different desired criteria an interpretation should satisfy, and focus on the faithfulness criteria. We survey the literature with respect to faithfulness evaluation, and arrange the current approaches around three assumptions, providing an explicit form to how faithfulness is "defined" by the community. We provide concrete guidelines on how evaluation of interpretation methods should and should not be conducted. Finally, we claim that the current binary definition for faithfulness sets a potentially unrealistic bar for being considered faithful. We call for discarding the binary notion of faithfulness in favor of a more graded one, which we believe will be of greater practical utility.

"How much of the Solar System should we reserve as wilderness, off-limits to human development?" We argue that, as a matter of policy, development should be limited to one eighth, with the remainder set aside. We argue that adopting a "1/8 principle" is far less restrictive, overall, than it might seem. One eighth of the iron in the asteroid belt is more than a million times greater than all of the Earth's estimated iron reserves and may suffice for centuries. A limit of some sort is needed because of the problems associated with exponential growth. Humans are poor at estimating the pace of such growth, so the limitations of a resource are hard to recognize before the final three doubling times which take utilization successively from 1/8 to 1/4 to 1/2, and then to the point of exhaustion. Population growth and climate change are instances of unchecked exponential growth. Each places strains upon ouru available resources. Each is a problem we would like to control but attempts to do so at this comparatively late stage have not been encouraging. Our limited ability to see ahead suggests that we should set ourselves a 'tripwire' that gives us at least 3 doubling times as leeway, i.e.

Model-based Reinforcement Learning (MBRL) holds promise for data-efficiency by planning with model-generated experience in addition to learning with experience from the environment. However, in complex or changing environments, models in MBRL will inevitably be imperfect, and their detrimental effects on learning can be difficult to mitigate. In this work, we question whether the objective of these models should be the accurate simulation of environment dynamics at all. We focus our investigations on Dyna-style planning in a prediction setting. First, we highlight and support three motivating points: a perfectly accurate model of environment dynamics is not practically achievable, is not necessary, and is not always the most useful anyways. Second, we introduce a meta-learning algorithm for training models with a focus on their usefulness to the learner instead of their accuracy in modelling the environment. Our experiments show that in a simple non-stationary environment, our algorithm enables faster learning than even using an accurate model built with domain-specific knowledge of the non-stationarity.

Standards govern the SHOULD and MUST requirements for protocol implementers for interoperability. In case of TCP that carries the bulk of the Internets' traffic, these requirements are defined in RFCs. While it is known that not all optional features are implemented and nonconformance exists, one would assume that TCP implementations at least conform to the minimum set of MUST requirements. In this paper, we use Internet-wide scans to show how Internet hosts and paths conform to these basic requirements. We uncover a non-negligible set of hosts and paths that do not adhere to even basic requirements. For example, we observe hosts that do not correctly handle checksums and cases of middlebox interference for TCP options. We identify hosts that drop packets when the urgent pointer is set or simply crash. Our publicly available results highlight that conformance to even fundamental protocol requirements should not be taken for granted but instead checked regularly.

Although several methods of environmental sound synthesis have been proposed, there has been no discussion on how synthesized environmental sounds should be evaluated. Only either subjective or objective evaluations have been conducted in conventional evaluations, and it is not clear what type of evaluation should be carried out. In this paper, we investigate how to evaluate synthesized environmental sounds. We also propose a subjective evaluation methodology to evaluate whether the synthesized sound appropriately represents the information input to the environmental sound synthesis system. In our experiments, we compare the proposed and conventional evaluation methods and show that the results of subjective evaluations tended to differ from those of objective evaluations. From these results, we conclude that it is necessary to conduct not only objective evaluation but also subjective evaluation.

Vector representations have become a central element in semantic language modelling, leading to mathematical overlaps with many fields including quantum theory. Compositionality is a core goal for such representations: given representations for 'wet' and 'fish', how should the concept 'wet fish' be represented? This position paper surveys this question from two points of view. The first considers the question of whether an explicit mathematical representation can be successful using only tools from within linear algebra, or whether other mathematical tools are needed. The second considers whether semantic vector composition should be explicitly described mathematically, or whether it can be a model-internal side-effect of training a neural network. A third and newer question is whether a compositional model can be implemented on a quantum computer. Given the fundamentally linear nature of quantum mechanics, we propose that these questions are related, and that this survey may help to highlight candidate operations for future quantum implementation.

In a paper by Alberto Caballero, a methodology is stated for estimating the probability of an exoplanet alien civilization having malicious intentions toward the human civilization after being messaged by it and estimating the number of malicious exoplanet civilizations in the Milky Way. Caballero states his paper attempts to provide these estimates. Caballero's methodology uses questionable hypotheses and excludes important parameters, indicating his methodology and estimates should be rejected.

The most common assumption in evolutionary game theory is that players should adopt a strategy that warrants the highest payoff. However, recent studies indicate that the spatial selection for cooperation is enhanced if an appropriate fraction of the population chooses the most common rather than the most profitable strategy within the interaction range. Such conformity might be due to herding instincts or crowd behavior in humans and social animals. In a heterogeneous population where individuals differ in their degree, collective influence, or other traits, an unanswered question remains who should conform. Selecting conformists randomly is the simplest choice, but it is neither a realistic nor the optimal one. We show that, regardless of the source of heterogeneity and game parametrization, socially the most favorable outcomes emerge if the masses conform. On the other hand, forcing leaders to conform significantly hinders the constructive interplay between heterogeneity and coordination, leading to evolutionary outcomes that are worse still than if conformists were chosen randomly. We conclude that leaders must be able to create a following for network reciprocity to be optimal

The quantities of dimension one - known as the dimensionless quantities - are widely used in physics. However, the debate about some dimensionless units is still open. The paper brings new interrelated arguments that lead to the conclusion to avoid physical dimensionless units, except one for the mathematical multiplication identity element that should not be introduced into a system of physical units. It brings the coherence to the International System of Units (SI) and it will remove ambiguities rising from the conflict between the mathematical properties and the physical conventions.

After four decades of research there still exists a Classification accuracy gap of about 20% between our best Unsupervisedly Learned Representations methods and the accuracy rates achieved by intelligent animals. It thus may well be that we are looking in the wrong direction. A possible solution to this puzzle is presented. We demonstrate that Reinforcement Learning can learn representations which achieve the same accuracy as that of animals. Our main modest contribution lies in the observations that: a. when applied to a real world environment Reinforcement Learning does not require labels, and thus may be legitimately considered as Unsupervised Learning, and b. in contrast, when Reinforcement Learning is applied in a simulated environment it does inherently require labels and should thus be generally be considered as Supervised Learning. The corollary of these observations is that further search for Unsupervised Learning competitive paradigms which may be trained in simulated environments may be futile.

Journal ranking is becoming more important in assessing the quality of academic research. Several indices have been suggested for this purpose, typically on the basis of a citation graph between the journals. We follow an axiomatic approach and find an impossibility theorem: any self-consistent ranking method, which satisfies a natural monotonicity property, should depend on the level of aggregation. Our result presents a trade-off between two axiomatic properties and reveals a dilemma of aggregation.

Many computer vision applications require solving multiple tasks in real-time. A neural network can be trained to solve multiple tasks simultaneously using multi-task learning. This can save computation at inference time as only a single network needs to be evaluated. Unfortunately, this often leads to inferior overall performance as task objectives can compete, which consequently poses the question: which tasks should and should not be learned together in one network when employing multi-task learning? We study task cooperation and competition in several different learning settings and propose a framework for assigning tasks to a few neural networks such that cooperating tasks are computed by the same neural network, while competing tasks are computed by different networks. Our framework offers a time-accuracy trade-off and can produce better accuracy using less inference time than not only a single large multi-task neural network but also many single-task networks.

In this paper, we study implications of the geometrical nature of space- time for some of the basic tenets of quantum mechanics. That is, we study two different implications of the principle of general covariance; first we quantize a reparametrization invariant theory, the free particle in Minkowski spacetime and point out in detail where this theory fails (no- tably these comments appear to be missing in the literature). Second we study the covariance of quantum field theory and show how it connects to causality, the outcome of this study is that QFT is what we shall call ultra weakly covariant with respect to the background spacetime. Third, we treat the question of whether evolution in quantum theory (apart from the measurement act) needs to be unitary, it is easily shown that a per- fectly satisfying probabilistic interpretation exists which does not require unitary evolution. Fourth, we speculate on some modifications quantum theory should undergo in order for it to be generally covariant. The results in this paper hint at a profound change of the theory in which causality as a fundamental principle is abandonned.

Elementary physical reasoning seems to leave it inevitable that global warming would increase the variability of the weather. The first two terms in an approximation to the global entropy are used to show that global warming has increased the free energy available to drive the weather, and that the variance of the weather should increase correspondingly.

There is a significant body of literature, which includes Itamar Pitowksy's "Betting on Outcomes of Measurements," that sheds light on the structure of quantum mechanics, and the ways in which it differs from classical mechanics, by casting the theory in terms of agents' bets on the outcomes of experiments. Though this approach, by itself, is neutral as to the ontological status of quantum observables and quantum states, some, notably those who adopt the label "QBism" for their views, take this approach as providing incentive to conclude that quantum states represent nothing in physical reality, but, rather, merely encode an agent's beliefs. In this chapter, I will argue that the arguments for realism about quantum states go through when the probabilities involved are taken to be subjective, if the conclusion is about the agent's beliefs: an agent whose credences conform to quantum probabilities should believe that preparation procedures with which she associates distinct pure quantum states produce distinct states of reality. The conclusion can be avoided only by stipulation of limitations on the agent's theorizing about the world, limitations that are not warranted by the empiric

Symplectic integrators are the preferred method of solving conservative $N$-body problems in cosmological, stellar cluster, and planetary system simulations because of their superior error properties and ability to compute orbital stability. Newtonian gravity is scale free, and there is no preferred time or length scale: this is at odds with construction of traditional symplectic integrators, in which there is an explicit timescale in the time-step. Additional timescales have been incorporated into symplectic integration using various techniques, such as hybrid methods and potential decompositions in planetary astrophysics, integrator sub-cycling in cosmology, and block time-stepping in stellar astrophysics, at the cost of breaking or potentially breaking symplecticity at a few points in phase space. The justification provided, if any, for this procedure is that these trouble points where the symplectic structure is broken should be rarely or never encountered in practice. We consider the case of hybrid integrators, which are used ubiquitously in astrophysics and other fields, to show that symplecticity breaks at a few points are sufficient to destroy beneficial properties of sympl

Shock/sound propagation from the quenched jets have well-defined front, separating the fireball into regions which are and are not affected. While even for the most robust jet quenching observed this increases local temperature and flow of ambient matter by only few percent at most, strong radial flow increases the contrast between the two regions so that the difference should be well seen in particle spectra at some $p_t$, perhaps even on event-by-event basis. We further show that the effect comes mostly from certain ellipse-shaped 1-d curve, the intercept of three 3-d surfaces, the Mach cone history, the timelike and spacelike freezeout surfaces. We further suggest that this "edge" is already seen in an event released by ATLAS collaboration.