The Eighty Five Percent Rule for optimal learning.
Ontology highlight
ABSTRACT:
SUBMITTER: Wilson RC
PROVIDER: S-EPMC6831579 | biostudies-literature | 2019 Nov
REPOSITORIES: biostudies-literature
Publications
The Eighty Five Percent Rule for optimal learning.
Wilson Robert C RC Shenhav Amitai A Straccia Mark M Cohen Jonathan D JD
Nature communications 20191105 1
Researchers and educators have long wrestled with the question of how best to teach their clients be they humans, non-human animals or machines. Here, we examine the role of a single variable, the difficulty of training, on the rate of learning. In many situations we find that there is a sweet spot in which training is neither too easy nor too hard, and where learning progresses most quickly. We derive conditions for this sweet spot for a broad class of learning algorithms in the context of bina ...[more]
Similar Datasets
Project description:This article studies the targeted sequential inference of an optimal treatment rule (TR) and its mean reward in the non-exceptional case, i.e., assuming that there is no stratum of the baseline covariates where treatment is neither beneficial nor harmful, and under a companion margin assumption. Our pivotal estimator, whose definition hinges on the targeted minimum loss estimation (TMLE) principle, actually infers the mean reward under the current estimate of the optimal TR. This data-adaptive statistical parameter is worthy of interest on its own. Our main result is a central limit theorem which enables the construction of confidence intervals on both mean rewards under the current estimate of the optimal TR and under the optimal TR itself. The asymptotic variance of the estimator takes the form of the variance of an efficient influence curve at a limiting distribution, allowing to discuss the efficiency of inference. As a by product, we also derive confidence intervals on two cumulated pseudo-regrets, a key notion in the study of bandits problems. A simulation study illustrates the procedure. One of the corner-stones of the theoretical study is a new maximal inequality for martingales with respect to the uniform entropy integral.
| S-EPMC5794253 | biostudies-literature
Project description:CO? has been long recognized for its anticonvulsant properties. We aimed to determine whether inhaling 5% CO? can be used to suppress seizures in epilepsy patients. The effect of CO? on cortical epileptic activity accompanying behavioral seizures was studied in rats and nonhuman primates, and based on these data, preliminary tests were carried out in humans.? In freely moving rats, cortical afterdischarges paralleled by myoclonic convulsions were evoked by sensorimotor cortex stimulation. Five percent CO? was applied for 5 min, 3 min before stimulation. In macaque monkeys, hypercarbia was induced by hypoventilation while seizure activity was electrically or chemically evoked in the sensorimotor cortex. Seven patients with drug-resistant partial epilepsy were examined with video-EEG (electroencephalography) and received 5% CO? in medical carbogen shortly after electrographic seizure onset.In rats, 5% CO? strongly suppressed cortical afterdischarges, by approximately 75%, whereas responses to single-pulse stimulation were reduced by about 15% only. In macaques, increasing pCO?) from 37 to 44-45 mm Hg (corresponding to inhalation of 5% CO? or less) suppressed stimulation-induced cortical afterdischarges by about 70% and single, bicuculline-induced epileptiform spikes by approximately 25%. In a pilot trial carried out in seven patients, a rapid termination of electrographic seizures was seen despite the fact that the application of 5% CO? was started after seizure generalization.Five percent CO? has a fast and potent anticonvulsant action. The present data suggest that medical carbogen with 5% CO? can be used for acute treatment to suppress seizures in epilepsy patients.
| S-EPMC3017646 | biostudies-literature
Project description:Objective:This bibliometric study investigated literature pertaining to a quickly growing population worldwide: the oldest-old, individuals age eighty-five and older. The current state of research was surveyed, based on top authors, publishers, authorship networks, themes in publication titles and abstracts, and highly cited publications. Methods:Bibliographic data was abstracted from the Web of Science database. Microsoft Excel was used for data analyses related to top author, publishers, and terms. VosViewer bibliographic visualization software was used to identify authorship networks. Results:Publications pertaining to the oldest-old have increased dramatically over the past three decades. The majority of these publications are related to medical or genetics topics. Citations for these publications remain relatively low but may be expected to grow in coming years, based on the publication behavior about and increasing prominence of this population. Claudio Franceschi and the Journal of the American Geriatrics Society were found to be the author and journal with the most publications pertaining to the oldest-old, respectively. Conclusions:The oldest-old is a population of rapidly growing significance. Researchers in library and information science, gerontology, and beyond can benefit themselves and those they serve by participating in research and specialized services to marginalized populations like the oldest-old. This bibliometric study hopefully serves as a launch-point for further inquiry and research in the years to come.
| S-EPMC6919997 | biostudies-literature
Project description:While the history of taxonomic diversification in open ocean lineages of ray-finned fish and elasmobranchs is increasingly known, the evolution of their roles within the open ocean ecosystem remains poorly understood. To assess the relative importance of these groups through time, we measured the accumulation rate of microfossil fish teeth and elasmobranch dermal denticles (ichthyoliths) in deep-sea sediment cores from the North and South Pacific gyres over the past 85 million years (Myr). We find three distinct and stable open ocean ecosystem structures, each defined by the relative and absolute abundance of elasmobranch and ray-finned fish remains. The Cretaceous Ocean (pre-66 Ma) was characterized by abundant elasmobranch denticles, but low abundances of fish teeth. The Palaeogene Ocean (66-20 Ma), initiated by the Cretaceous/Palaeogene mass extinction, had nearly four times the abundance of fish teeth compared with elasmobranch denticles. This Palaeogene Ocean structure remained stable during the Eocene greenhouse (50 Ma) and the Eocene-Oligocene glaciation (34 Ma), despite large changes in the overall accumulation of both groups during those intervals, suggesting that climate change is not a primary driver of ecosystem structure. Dermal denticles virtually disappeared from open ocean ichthyolith assemblages approximately 20 Ma, while fish tooth accumulation increased dramatically in variability, marking the beginning of the Modern Ocean. Together, these results suggest that open ocean fish community structure is stable on long timescales, independent of total production and climate change. The timing of the abrupt transitions between these states suggests that the transitions may be due to interactions with other, non-preserved pelagic consumer groups.
| S-EPMC4892790 | biostudies-literature
Project description:In order to improve the discovery and development of new drugs, a broad effort is being made to assess the 'drug-like' properties of molecules in early stages of the discovery-research process. Although there are numerous approaches to this problem, perhaps the simplest and most widespread one is that developed by Chris Lipinski and his co-workers at Pfizer, which is generally referred either as the Lipinski Rules or the Rule of Five (ROF). The ROF is based on four properties of molecules, namely, molecular weight (MW), log P, number of hydrogen bond donors (HBD), and the number of hydrogen bond acceptors (HBA). A 'flag' is set if the value of a given property exceeds the chosen threshold value for that property-MW 500 Da, log P 5, the number of HBDs 5, and the number of HBAs 10. Each flag corresponds to an ROF violation. The total number of violations is the ROF-Score, which lies between '0' and '4'. Molecules with ROF-Scores greater than one are considered to be marginal for further development. The difficulty with this approach is that two molecules with nearly identical property values can, nonetheless, possess ROF-Scores that can differ by two or more. Thus, one molecule could be considered for further studies while the other, nearly identical molecule (in terms of its four ROF properties), would most likely not be. This problem arises because of the sharp thresholds imposed by the present formulation of the ROF, which is based upon classical sets. In the current work an alternative approach based on the use of utility functions, within the framework of the analytic hierarchy process (AHP), are employed to 'soften' the sharp boundaries inherent in classical sets. This provides a more realistic assessment of compounds in terms of their potential suitability in drug-discovery research programs.
| S-EPMC4209914 | biostudies-literature
Project description:Beyond rule-of-five (bRo5) compounds are increasingly used in drug discovery. Here we analyze 37 target proteins that have bRo5 drugs or clinical candidates. Targets can benefit from bRo5 drugs if they have "complex" hot spot structure with four or more hots spots, including some strong ones. Complex I targets show positive correlation between binding affinity and molecular weight. These targets are conventionally druggable, but reaching additional hot spots enables improved pharmaceutical properties. Complex II targets, mostly protein kinases, also have strong hot spots but show no correlation between affinity and ligand molecular weight, and the primary motivation for creating larger drugs is to increase selectivity. Each target considered as complex III has some specific reason for requiring bRo5 drugs. Finally, targets with "simple" hot spot structure, i.e., three or fewer weak hot spots, must use larger compounds that interact with surfaces beyond the hot spot region to achieve acceptable affinity.
| S-EPMC7102492 | biostudies-literature
Project description:We developed a biologically plausible unsupervised learning algorithm, error-gated Hebbian rule (EGHR)-?, that performs principal component analysis (PCA) and independent component analysis (ICA) in a single-layer feedforward neural network. If parameter ??=?1, it can extract the subspace that major principal components span similarly to Oja's subspace rule for PCA. If ??=?0, it can separate independent sources similarly to Bell-Sejnowski's ICA rule but without requiring the same number of input and output neurons. Unlike these engineering rules, the EGHR-? can be easily implemented in a biological or neuromorphic circuit because it only uses local information available at each synapse. We analytically and numerically demonstrate the reliability of the EGHR-? in extracting and separating major sources given high-dimensional input. By adjusting ?, the EGHR-? can extract sources that are missed by the conventional engineering approach that first applies PCA and then ICA. Namely, the proposed rule can successfully extract hidden natural images even in the presence of dominant or non-Gaussian noise components. The results highlight the reliability and utility of the EGHR-? for large-scale parallel computation of PCA and ICA and its future implementation in a neuromorphic hardware.
| S-EPMC5789861 | biostudies-other
Project description:Myxofibrosarcoma and undifferentiated pleomorphic sarcoma (UPS) are aggressive, genetically complex sarcomas. The minimum myxoid component used as a criterion for myxofibrosarcoma varies widely, so we determined the optimal myxoid component cutpoints for stratifying outcomes of UPS and myxofibrosarcoma. We also analyzed clinicopathologic factors associated with outcome.Review of a prospective, single-institution database identified 197 patients with primary, high-grade extremity/truncal myxofibrosarcoma or UPS resected during 1992-2013. Histology was reviewed and percent myxoid component determined for each tumor. Disease-specific survival (DSS) and distant recurrence-free survival (DRFS) were analyzed using the Kaplan-Meier method, log-rank test, and Cox regression.Median follow-up for survivors was 6.4 years. In minimum p value analysis of myxoid component, the best cutpoint for both DSS and DRFS was 5% (adjusted p ? 0.001), followed by 70%. Therefore, sarcomas with <5% myxoid component (n = 69) were classified as UPS and those with ?5% myxoid component (n = 128) as myxofibrosarcoma. Five-year DRFS was 24% for UPS, 51% for 5-69% myxoid component myxofibrosarcoma, and 65% for ?70% myxoid component myxofibrosarcoma. Myxoid component, tumor size, and age were independently associated with DSS; myxoid component and tumor size were associated with DRFS. Only tumor site was associated with local recurrence.Percent myxoid component and tumor size are the two most important predictors of DSS and DRFS in high-grade myxofibrosarcoma and UPS. A 5% myxoid component cutpoint is an improved criterion for classifying myxofibrosarcoma. Myxoid component-based classification improves stratification of patient outcome and will aid in selection of patients for systemic therapy and clinical trials.
| S-EPMC4964786 | biostudies-literature
Project description:MotivationDisease state prediction from biomarker profiling studies is an important problem because more accurate classification models will potentially lead to the discovery of better, more discriminative markers. Data mining methods are routinely applied to such analyses of biomedical datasets generated from high-throughput 'omic' technologies applied to clinical samples from tissues or bodily fluids. Past work has demonstrated that rule models can be successfully applied to this problem, since they can produce understandable models that facilitate review of discriminative biomarkers by biomedical scientists. While many rule-based methods produce rules that make predictions under uncertainty, they typically do not quantify the uncertainty in the validity of the rule itself. This article describes an approach that uses a Bayesian score to evaluate rule models.ResultsWe have combined the expressiveness of rules with the mathematical rigor of Bayesian networks (BNs) to develop and evaluate a Bayesian rule learning (BRL) system. This system utilizes a novel variant of the K2 algorithm for building BNs from the training data to provide probabilistic scores for IF-antecedent-THEN-consequent rules using heuristic best-first search. We then apply rule-based inference to evaluate the learned models during 10-fold cross-validation performed two times. The BRL system is evaluated on 24 published 'omic' datasets, and on average it performs on par or better than other readily available rule learning methods. Moreover, BRL produces models that contain on average 70% fewer variables, which means that the biomarker panels for disease prediction contain fewer markers for further verification and validation by bench scientists.
| S-EPMC2852212 | biostudies-literature
Project description:Humans and animals can flexibly switch rules to generate the appropriate response to the same sensory stimulus, e.g., we kick a soccer ball toward a friend on our team, but we kick the ball away from a friend who is traded to an opposing team. Most motor learning experiments have relied on a fixed rule; therefore, the effects of switching rules on motor learning are unclear. Here, we study the availability of motor learning effects when a rule in the training phase is different from a rule in the probe phase. Our results suggest that switching a rule causes partial rather than perfect availability. To understand the neural mechanisms inherent in our results, we verify that a computational model can explain our experimental results when each neural unit has different activities, but the total population activity is the same in the same planned movement with different rules. Thus, we conclude that switching rules causes modulations in individual neural activities under the same population activity, resulting in a partial transfer of learning effects for the same planned movements. Our results indicate that sports training and rehabilitation should include various situations even when the same motions are required.
| S-EPMC6131250 | biostudies-literature