The concept of free energy has its origins in 19th century thermodynamics, but has recently found its way into the behavioral and neural sciences, where it has been promoted for its wide ...applicability and has even been suggested as a fundamental principle of understanding intelligent behavior and brain function. We argue that there are essentially two different notions of free energy in current models of intelligent agency, that can both be considered as applications of Bayesian inference to the problem of action selection: one that appears when trading off accuracy and uncertainty based on a general maximum entropy principle, and one that formulates action selection in terms of minimizing an error measure that quantifies deviations of beliefs and policies from given reference models. The first approach provides a normative rule for action selection in the face of model uncertainty or when information processing capabilities are limited. The second approach directly aims to formulate the action selection problem as an inference problem in the context of Bayesian brain theories, also known as Active Inference in the literature. We elucidate the main ideas and discuss critical technical and conceptual issues revolving around these two notions of free energy that both claim to apply at all levels of decision-making, from the high-level deliberation of reasoning down to the low-level information processing of perception.
Celotno besedilo
Dostopno za:
DOBA, IZUM, KILJ, NUK, PILJ, PNG, SAZU, SIK, UILJ, UKNU, UL, UM, UPUK
In its most basic form, decision-making can be viewed as a computational process that progressively eliminates alternatives, thereby reducing uncertainty. Such processes are generally costly, meaning ...that the amount of uncertainty that can be reduced is limited by the amount of available computational resources. Here, we introduce the notion of elementary computation based on a fundamental principle for probability transfers that reduce uncertainty. Elementary computations can be considered as the inverse of Pigou-Dalton transfers applied to probability distributions, closely related to the concepts of majorization, T-transforms, and generalized entropies that induce a preorder on the space of probability distributions. Consequently, we can define resource cost functions that are order-preserving and therefore monotonic with respect to the uncertainty reduction. This leads to a comprehensive notion of decision-making processes with limited resources. Along the way, we prove several new results on majorization theory, as well as on entropy and divergence measures.
The tribological behavior of steel sliding pairs with dimples ranging from 15 to 800µm in diameter was characterized in the mixed lubrication regime in a pin-on-disk experiment. The pin was textured, ...keeping the total dimpled area constant at 10% and the depth-to-diameter ratio at 0.1. Polyalphaolefin (PAO) was used as a model lubricant. Experiments were carried out under unidirectional sliding conditions at 50 and 100°C. At constant depth-to-diameter ratio the results showed a significant non-linear dependence of the friction coefficient on the texture diameter, sliding speed and the oil temperature (viscosity). A friction reduction of up to 80% was possible with the optimal diameter for certain sliding speeds. The dimple diameters leading to the highest friction reduction significantly depend on the oil temperature. By reducing the oil temperature from 100 to 50°C the dimple diameter resulting in the highest friction reduction changed from 40 to 200µm.
•Over 80% of friction reduction by laser surface texturing under mixed lubrication.•The friction reduction depends on the dimple diameter.•Two competing mechanisms: (a) the number of dimples edges, and (b) the dimple size.•The optimum dimple diameter depends on the oil temperature.
Quantum metrology promises high-precision measurements of classical parameters with far reaching implications for science and technology. So far, research has concentrated almost exclusively on ...quantum-enhancements in integrable systems, such as precessing spins or harmonic oscillators prepared in non-classical states. Here we show that large benefits can be drawn from rendering integrable quantum sensors chaotic, both in terms of achievable sensitivity as well as robustness to noise, while avoiding the challenge of preparing and protecting large-scale entanglement. We apply the method to spin-precession magnetometry and show in particular that the sensitivity of state-of-the-art magnetometers can be further enhanced by subjecting the spin-precession to non-linear kicks that renders the dynamics chaotic.
One notable weakness of current machine learning algorithms is the poor ability of models to solve new problems without forgetting previously acquired knowledge. The Continual Learning paradigm has ...emerged as a protocol to systematically investigate settings where the model sequentially observes samples generated by a series of tasks. In this work, we take a task-agnostic view of continual learning and develop a hierarchical information-theoretic optimality principle that facilitates a trade-off between learning and forgetting. We derive this principle from a Bayesian perspective and show its connections to previous approaches to continual learning. Based on this principle, we propose a neural network layer, called the Mixture-of-Variational-Experts layer, that alleviates forgetting by creating a set of information processing paths through the network which is governed by a gating policy. Equipped with a diverse and specialized set of parameters, each path can be regarded as a distinct sub-network that learns to solve tasks. To improve expert allocation, we introduce diversity objectives, which we evaluate in additional ablation studies. Importantly, our approach can operate in a task-agnostic way, i.e., it does not require task-specific knowledge, as is the case with many existing continual learning algorithms. Due to the general formulation based on generic utility functions, we can apply this optimality principle to a large variety of learning problems, including supervised learning, reinforcement learning, and generative modeling. We demonstrate the competitive performance of our method on continual reinforcement learning and variants of the MNIST, CIFAR-10, and CIFAR-100 datasets.
Summary Background Ixekizumab is a humanised monoclonal antibody against the proinflammatory cytokine interleukin 17A. We report two studies of ixekizumab compared with placebo or etanercept to ...assess the safety and efficacy of specifically targeting interleukin 17A in patients with widespread moderate-to-severe psoriasis. Methods In two prospective, double-blind, multicentre, phase 3 studies (UNCOVER-2 and UNCOVER-3), eligible patients were aged 18 years or older, had a confirmed diagnosis of chronic plaque psoriasis at least 6 months before baseline (randomisation), 10% or greater body-surface area involvement at both screening and baseline visits, at least a moderate clinical severity as measured by a static physician global assessment (sPGA) score of 3 or more, and a psoriasis area and severity index (PASI) score of 12. Participants were randomly assigned (1:2:2:2) by computer-generated random sequence with an interactive voice response system to receive subcutaneous placebo, etanercept (50 mg twice weekly), or one injection of 80 mg ixekizumab every 2 weeks, or every 4 weeks after a 160 mg starting dose. Blinding was maintained with a double-dummy design. Coprimary efficacy endpoints were proportions of patients achieving sPGA score 0 or 1 and 75% or greater improvement in PASI at week 12. Analysis was by intention to treat. These trials are registered with ClinicalTrials.gov , numbers NCT01597245 and NCT01646177. Findings Between May 30, 2012, and Dec 30, 2013, 1224 patients in UNCOVER-2 were randomly assigned to receive subcutaneous placebo (n=168), etanercept (n=358), or ixekizumab every 2 weeks (n=351) or every 4 weeks (n=347); between Aug 11, 2012, and Feb 27, 2014, 1346 patients in UNCOVER-3 were randomly assigned to receive placebo (n=193), etanercept (n=382), ixekizumab every 2 weeks (n=385), or ixekizumab every 4 weeks (n=386). At week 12, both primary endpoints were met in both studies. For UNCOVER-2 and UNCOVER-3 respectively, in the ixekizumab every 2 weeks group, PASI 75 was achieved by 315 (response rate 89·7%; effect size 87·4% (97·5% CI 82·9–91·8) vs placebo; 48·1% (41·2–55·0) vs etanercept) and 336 (87·3%; 80·0% (74·4–85·7) vs placebo; 33·9% (27·0–40·7) vs etanercept) patients; in the ixekizumab every 4 weeks group, by 269 (77·5%; 75·1% (69·5–80·8) vs placebo; 35·9% (28·2–43·6) vs etanercept) and 325 (84·2%; 76·9% (71·0–82·8) vs placebo; 30·8% (23·7–37·9) vs etanercept) patients; in the placebo group, by four (2·4%) and 14 (7·3%) patients; and in the etanercept group by 149 (41·6%) and 204 (53·4%) patients (all p<0·0001 vs placebo or etanercept). In the ixekizumab every 2 weeks group, sPGA 0/1 was achieved by 292 (response rate 83·2%; effect size 80·8% (97·5% CI 75·6–86·0) vs placebo; 47·2% (39·9–54·4) vs etanercept) and 310 (80·5%; 73·8% (67·7–79·9) vs placebo; 38·9% (31·7–46·1) vs etanercept) patients; in the ixekizumab every 4 weeks group by 253 (72·9%; 70·5% (64·6–76·5) vs placebo; 36·9% (29·1–44·7) vs etanercept) and 291 (75·4%; 68·7% (62·3–75·0) vs placebo; 33·8% (26·3–41·3) vs etanercept) patients; in the placebo group by four (2·4%) and 13 (6·7%) patients; and in the etanercept group by 129 (36·0%) and 159 (41·6%) patients (all p<0·0001 vs placebo or etanercept). In combined studies, serious adverse events were reported in 14 (1·9%) of 734 patients given ixekizumab every 2 weeks, 14 (1·9%) of 729 given ixekizumab every 4 weeks, seven (1·9%) of 360 given placebo, and 14 (1·9%) of 739 given etanercept; no deaths were noted. Interpretation Both ixekizumab dose regimens had greater efficacy than placebo and etanercept over 12 weeks in two independent studies. These studies show that selectively neutralising interleukin 17A with a high affinity antibody potentially gives patients with psoriasis a new and effective biological therapy option. Funding Eli Lilly and Co.
In three phase 3 trials, ixekizumab, an anti–IL-17A monoclonal antibody, was effective in the treatment of patients with moderate-to-severe plaque psoriasis. Adverse events included neutropenia, ...candida infections, and inflammatory bowel disease.
Psoriasis is a chronic inflammatory disease that is mediated by aberrant immune responses and driven by self-perpetuating cytokine networks.
1
,
2
Advances in understanding the pathogenic cytokine network of psoriasis have led to the development of new treatments
3
–
5
that provide greater efficacy in terms of complete skin clearance.
6
–
9
The motivation to completely clear psoriasis plaques from the skin of patients has grown in response to accumulating evidence that residual skin disease can affect a patient’s health-related quality of life
10
–
12
similar to that associated with chronic conditions such as type 2 diabetes.
13
Ixekizumab, a recombinant, high-affinity, humanized, IgG4-κ monoclonal . . .
Perfectly rational decision-makers maximize expected utility, but crucially ignore the resource costs incurred when determining optimal actions. Here, we propose a thermodynamically inspired ...formalization of bounded rational decision-making where information processing is modelled as state changes in thermodynamic systems that can be quantified by differences in free energy. By optimizing a free energy, bounded rational decision-makers trade off expected utility gains and information-processing costs measured by the relative entropy. As a result, the bounded rational decision-making problem can be rephrased in terms of well-known variational principles from statistical physics. In the limit when computational costs are ignored, the maximum expected utility principle is recovered. We discuss links to existing decision-making frameworks and applications to human decision-making experiments that are at odds with expected utility theory. Since most of the mathematical machinery can be borrowed from statistical physics, the main contribution is to re-interpret the formalism of thermodynamic free-energy differences in terms of bounded rational decision-making and to discuss its relationship to human decision-making experiments.
Recently proposed quantum-chaotic sensors achieve quantum enhancements in measurement precision by applying nonlinear control pulses to the dynamics of the quantum sensor while using classical ...initial states that are easy to prepare. Here, we use the cross-entropy method of reinforcement learning (RL) to optimize the strength and position of control pulses. Compared to the quantum-chaotic sensors with periodic control pulses in the presence of superradiant damping, we find that decoherence can be fought even better and measurement precision can be enhanced further by optimizing the control. In some examples, we find enhancements in sensitivity by more than an order of magnitude. By visualizing the evolution of the quantum state, the mechanism exploited by the RL method is identified as a kind of spin-squeezing strategy that is adapted to the superradiant damping.
Rate distortion theory describes how to communicate relevant information most efficiently over a channel with limited capacity. One of the many applications of rate distortion theory is bounded ...rational decision making, where decision makers are modeled as information channels that transform sensory input into motor output under the constraint that their channel capacity is limited. Such a bounded rational decision maker can be thought to optimize an objective function that trades off the decision maker’s utility or cumulative reward against the information processing cost measured by the mutual information between sensory input and motor output. In this study, we interpret a spiking neuron as a bounded rational decision maker that aims to maximize its expected reward under the computational constraint that the mutual information between the neuron’s input and output is upper bounded. This abstract computational constraint translates into a penalization of the deviation between the neuron’s instantaneous and average firing behavior. We derive a synaptic weight update rule for such a rate distortion optimizing neuron and show in simulations that the neuron efficiently extracts reward-relevant information from the input by trading off its synaptic strengths against the collected reward.
Celotno besedilo
Dostopno za:
DOBA, IZUM, KILJ, NUK, PILJ, PNG, SAZU, SIK, UILJ, UKNU, UL, UM, UPUK
PDF
