Filters








1,231 Hits in 3.3 sec

Acceleration via Fractal Learning Rate Schedules [article]

Naman Agarwal, Surbhi Goel, Cyril Zhang
2021 arXiv   pre-print
We provide some experiments to challenge conventional beliefs about stable learning rates in deep learning: the fractal schedule enables training to converge with locally unstable updates which make negative  ...  We reinterpret an iterative algorithm from the numerical analysis literature as what we call the Chebyshev learning rate schedule for accelerating vanilla gradient descent, and show that the problem of  ...  A fractal Chebyshev schedule with m = 1/20, M = 1, T = 64 accelerated convergence when applied to the stable constant learning rate baseline.  ... 
arXiv:2103.01338v2 fatcat:q34e74ifyfdaxhb4av4riol22m

Post-training deep neural network pruning via layer-wise calibration [article]

Ivan Lazarevich and Alexander Kozlov and Nikita Malinin
2021 arXiv   pre-print
We propose a data-free extension of the approach for computer vision models based on automatically-generated synthetic fractal images.  ...  We obtain state-of-the-art results for data-free neural network pruning, with ~1.5% top@1 accuracy drop for a ResNet50 on ImageNet at 50% sparsity rate.  ...  for the best configuration via global non-gradient optimization or reinforcement learning [7] .  ... 
arXiv:2104.15023v1 fatcat:o67pulxvsncnloartcg6d7uidi

Evidence for model-based encoding of Pavlovian contingencies in the human brain

Wolfgang M. Pauli, Giovanni Gentile, Sven Collette, Julian M. Tyszka, John P. O'Doherty
2019 Nature Communications  
However, these model-free approximations fall short of comprehensively capturing learning and behavior in Pavlovian conditioning.  ...  Prominent accounts of Pavlovian conditioning successfully approximate the frequency and intensity of conditioned responses under the assumption that learning is exclusively model-free; that animals do  ...  These tables also provide results for alternative training/testing schedules of classifiers across proximal and distal CS fractals.  ... 
doi:10.1038/s41467-019-08922-7 pmid:30846685 pmcid:PMC6405831 fatcat:qbpei2uyvvawblgl5vi3pjvua4

The dynamics of behavior (Editorial)

Gregory Galbicka
1992 Journal of The Experimental Analysis of Behavior  
It is not too surprising that correlated with the increasing popularity of VI schedules was a negatively accelerated frequency of cumulative records published in the journal.  ...  Linear-IRT schedules (Galbicka & Platt, 1984; Platt, 1979; Weiss, 1970) , cyclic-interval schedules (Innis & Staddon, 1971; McDowell & Sulzen, 1981; Staddon, 1964) , percentile schedules (Arbuckle &  ... 
doi:10.1901/jeab.1992.57-243 pmid:16812654 pmcid:PMC1323228 fatcat:4cowbmhzcrfxdjw3pbwipdvw5y

System-Cluster Technology of e Learning Improvement under the Conditions of COVID-19

Tsvetana Stoyanova, Philip Stoyanov, Anzhelika Remnova, Svitlana Kushniruk, Lyudmyla Rakityanska, Svetlana Drobyazko
2021 Sustainability  
The fractal-cluster technology of an e-learning organization was suggested for an introduction.  ...  The expediency of introducing fractal-cluster structures into the organizational component of the educational process was determined.  ...  Clusters Didactic Characteristics of Learning Complexes 1. Energetic cluster The rate of the learning information provision 2.  ... 
doi:10.3390/su132414024 fatcat:uxh4i7eiwnecpiodxqp36cxllu

Self-Awareness as a Model for Designing and Operating Heterogeneous Multicores

Andreas Agne, Markus Happe, Achim Lösch, Christian Plessl, Marco Platzner
2014 ACM Transactions on Reconfigurable Technology and Systems  
quality of service requirements but also to internal changes such as thermal problems and failures, and even anticipate such changes through modeling of the system and environment and through online learning  ...  We vary the rate to mimic a fractal workload W s exhibiting a degree of self-similarity that is commonly observed in a number of application domains, such as networking [Leland et al. 1994 ].  ...  Figure 4 (a) displays the modulated fractal sorting workload W s in blocks per second (BPS).  ... 
doi:10.1145/2617596 fatcat:ep5qzii7xzfofp5yc74m7vkk7u

Fractality of sensations and the brain health: the theory linking neurodegenerative disorder with distortion of spatial and temporal scale-invariance and fractal complexity of the visible world

Marina V. Zueva
2015 Frontiers in Aging Neuroscience  
Citation: Zueva MV (2015) Fractality of sensations and the brain health: the theory linking neurodegenerative disorder with distortion of spatial and temporal scale-invariance and fractal complexity of  ...  The deficit of fractal complexity of environmental influences can lead to the distortion of fractal complexity in the visual pathways of the brain and abnormalities of development or aging.  ...  The authors note that reduced schedule of brain activity, noisy processing, weakened neuromodulatory control, and negative learning -all promote plastic changes in the brain and functional decline.  ... 
doi:10.3389/fnagi.2015.00135 pmid:26236232 pmcid:PMC4502359 fatcat:mggs2vevffgu3isk3tipx6dtzy

Stress-induced impairment in goal-directed instrumental behaviour is moderated by baseline working memory

C.W.E.M. Quaedflieg, H. Stoffregen, I. Sebalo, T. Smeets
2019 Neurobiology of Learning and Memory  
To this end, 112 healthy participants performed an instrumental learning task. In phase 1, participants learned instrumental actions that were associated with two different food rewards.  ...  Moreover, enhanced dopaminergic activity during learning accelerated the transition from goal-directed to habitual performance in rats (Wickens, Horvitz, Costa, & Killcross, 2007) .  ...  Food outcomes were available on a variable interval schedule with an average of one outcome per ten seconds (VI-10).  ... 
doi:10.1016/j.nlm.2019.01.010 pmid:30664942 fatcat:cs76zr46pvc4pg4putz4rxkhbq

Strategies and Rubrics for Teaching Chaos and Complex Systems Theories as Elaborating, Self-Organizing, and Fractionating Evolutionary Systems

Lynn S. Fichter, E. J. Pyle, S. J. Whitmeyer
2010 Journal of Geoscience education  
through positive and negative feedbacks, forming interdependent, dynamic, evolutionary networks, that possess universality properties common to all complex systems (bifurcations, sensitive dependence, fractal  ...  We present a learning progression of concept building from chaos theory, through a variety of complex systems, and ending with how such systems result in increases in complexity, diversity, order, and/  ...  Anticipated Learning Outcomes: 10. All complex systems accelerate their rate of changebifurcations-at the same rate.  ... 
doi:10.5408/1.3534849 fatcat:sisxg4linjgclne4xmq6niel44

Neuroimaging of the Philadelphia Neurodevelopmental Cohort

Theodore D. Satterthwaite, Mark A. Elliott, Kosha Ruparel, James Loughead, Karthik Prabhakaran, Monica E. Calkins, Ryan Hopson, Chad Jackson, Jack Keefe, Marisa Riley, Frank D. Mentch, Patrick Sleiman (+5 others)
2014 NeuroImage  
This restructuring significantly increased the number of subjects scheduled and also lowered the no-show rate.  ...  Of the 1409 subjects scheduled for MRI as part of the revised recruitment strategy, only 16% did not arrive for their scheduled appointment, representing a nearly 50% decline in the no-show rate.  ... 
doi:10.1016/j.neuroimage.2013.07.064 pmid:23921101 pmcid:PMC3947233 fatcat:mm6jehpoarctvl3e2wzhr3jlam

Differential, but not opponent, effects of l-DOPA and citalopram on action learning with reward and punishment

Marc Guitart-Masip, Marcos Economides, Quentin J. M. Huys, Michael J. Frank, Rumana Chowdhury, Emrah Duzel, Peter Dayan, Raymond J. Dolan
2013 Psychopharmacology  
This results in a disadvantage in learning to go to avoid punishment and in learning to no-go to obtain a reward.  ...  Objective To investigate the role of dopamine and serotonin in the interaction between action and valence during learning.  ...  rate.  ... 
doi:10.1007/s00213-013-3313-4 pmid:24232442 pmcid:PMC3923110 fatcat:yuuoplshbrd77o3pwpudm6okq4

Shifting the balance between goals and habits: Five failures in experimental habit induction

Sanne de Wit, Merel Kindt, Sarah L. Knot, Aukje A. C. Verhoeven, Trevor W. Robbins, Julia Gasull-Camos, Michael Evans, Hira Mirza, Claire M. Gillan
2018 Journal of experimental psychology. General  
Specifically, it converges with the suggestion that the failures in outcome devaluation in compulsive individuals reflect dysfunction in goal-directed control, rather than overactive habit learning.  ...  Extensive training did not lead to greater habits in two versions of an avoidance learning task, in an appetitive slips-of-action task, or in two independent attempts to replicate the original demonstration  ...  First, a variable interval (VI) schedule was employed as opposed to a fixed ratio schedule.  ... 
doi:10.1037/xge0000402 pmid:29975092 pmcid:PMC6033090 fatcat:zemk3j4dcvfpxiujtgur3ri54e

Mouse Activity across Time Scales: Fractal Scenarios

G. Z. dos Santos Lima, B. Lobão-Soares, G. C. do Nascimento, Arthur S. C. França, L. Muratori, S. Ribeiro, G. Corso, Clayton T. Dickson
2014 PLoS ONE  
Finally, we compare scaling properties of body acceleration fluctuation time series during sleep and wake periods for healthy mice.  ...  In a similar way, an intense cardiac activity should be succeeded, via feedback control, by a diminution in the heart rate and vice-versa [7] .  ...  The acceleration is the variation of velocity, a particle that moves at constant velocity shows zero acceleration. Acceleration, in a crude way, is related with force.  ... 
doi:10.1371/journal.pone.0105092 pmid:25275515 pmcid:PMC4183474 fatcat:pbh5fedw6jairdj4rgjlca7tsy

Model Slicing for Supporting Complex Analytics with Elastic Inference Cost and Resource Constraints [article]

Shaofeng Cai, Gang Chen, Beng Chin Ooi, Jinyang Gao
2019 arXiv   pre-print
Supporting such deep learning service with dynamic workload cost-efficiently remains to be a challenging problem.  ...  structural constraint, we can slice a narrower sub-model during inference whose run-time memory and computational operation consumption is roughly quadratic to the width controlled by a single parameter slice rate  ...  For example, the error rate is 8.62% for VGG-13-0.375 (Table 3) when slice rate is 0.375, which costs only 14.06% of the computational operation of the full network (∼7.11x computation acceleration).  ... 
arXiv:1904.01831v1 fatcat:gvuyexnx3rgabixtx4fz46x6uu

Interdisciplinary Methodology to Extend Technology Readiness Levels in SONAR Simulation from Laboratory Validation to Hydrography Demonstrator

James Riordan, Francis Flannery, Daniel Toal, Matija Rossi, Gerard Dooly
2019 Journal of Marine Science and Engineering  
By providing access to the actual simulator via a cloud service, the learning can be transformed into an on-demand experience.  ...  Instructors can schedule on demand formative competency assessments without having to interleave with higher priority ship operations, while students have similar flexibility in scheduling their self-learning  ... 
doi:10.3390/jmse7050159 fatcat:mbehnhm7ina65jiglyftpxbe64
« Previous Showing results 1 — 15 out of 1,231 results