High-level features for resource economy and fast learning in skill transfer

Mathematical competence is one of the basic competences defined in the EU. Results of international studies in recent years show that the percentage of pupils in Latvia with high level (5.,6.) of mathematical competence has decreased from 8% (PISA, 2012) to 5,2% (PISA, 2015). Observations of mathematical lessons show that individualization of studies is not a popular everyday feature, nonstandard problems are rarely used in the work with primary school pupils. Sustainable education can not be envisioned without creative thinking necessary for solving various nonstandard problems. Mathematical competitions also require creative applications of knowledge. The goal of this study was to analyze problems of Latvian mathematical contests for grades 4-6 of the last 3 years according to categories of mathematical content. The most important cognitive and metacognitive strategies necessary for their solution are shown. It is important to turn attention of teachers to much wider inclusion of contest problems into study process of primary school. It will enable to individualize studies and stimulate skill transfer to new situations for gifted pupils. The authors encourage teachers to use nonstandard (contest) problems as an indivisualization tool which will give opportunity for pupils to master knowledge and skill transfer. It will provide regular training of mind and positive emotions for pupils who are bored with solving standard problems.

Download Full-text

Fast Learning of Simple Perceptual Discriminations Reduces Brain Activation in Working Memory and in High-level Auditory Regions

Journal of Cognitive Neuroscience ◽

10.1162/jocn_a_00786 ◽

2015 ◽

Vol 27 (7) ◽

pp. 1308-1321 ◽

Cited By ~ 9

Author(s):

Luba Daikhin ◽

Merav Ahissar

Keyword(s):

Working Memory ◽

Brain Activity ◽

Brain Activation ◽

Frequency Discrimination ◽

Reference Condition ◽

Tone Frequency ◽

Temporal Area ◽

Fast Learning ◽

High Level ◽

Online Storage

Introducing simple stimulus regularities facilitates learning of both simple and complex tasks. This facilitation may reflect an implicit change in the strategies used to solve the task when successful predictions regarding incoming stimuli can be formed. We studied the modifications in brain activity associated with fast perceptual learning based on regularity detection. We administered a two-tone frequency discrimination task and measured brain activation (fMRI) under two conditions: with and without a repeated reference tone. Although participants could not explicitly tell the difference between these two conditions, the introduced regularity affected both performance and the pattern of brain activation. The “No-Reference” condition induced a larger activation in frontoparietal areas known to be part of the working memory network. However, only the condition with a reference showed fast learning, which was accompanied by a reduction of activity in two regions: the left intraparietal area, involved in stimulus retention, and the posterior superior-temporal area, involved in representing auditory regularities. We propose that this joint reduction reflects a reduction in the need for online storage of the compared tones. We further suggest that this change reflects an implicit strategic shift “backwards” from reliance mainly on working memory networks in the “No-Reference” condition to increased reliance on detected regularities stored in high-level auditory networks.

Download Full-text

Independent Skill Transfer for Deep Reinforcement Learning

Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence ◽

10.24963/ijcai.2020/401 ◽

2020 ◽

Author(s):

Qiangxing Tian ◽

Guanchu Wang ◽

Jinxin Liu ◽

Donglin Wang ◽

Yachen Kang

Keyword(s):

Reinforcement Learning ◽

High Efficiency ◽

Skill Transfer ◽

Lower Dimension ◽

Independent Components ◽

Intrinsic Reward ◽

Low Efficiency ◽

Transfer Method ◽

High Level ◽

Robotic Tasks

Recently, diverse primitive skills have been learned by adopting the entropy as intrinsic reward, which further shows that new practical skills can be produced by combining a variety of primitive skills. This is essentially skill transfer, very useful for learning high-level skills but quite challenging due to the low efficiency of transferring primitive skills. In this paper, we propose a novel efficient skill transfer method, where we learn independent skills and only independent components of skills are transferred instead of the whole set of skills. More concretely, independent components of skills are obtained through independent component analysis (ICA), which always have a smaller amount (or lower dimension) compared with their mixtures. With a lower dimension, independent skill transfer (IST) exhibits a higher efficiency on learning a given task. Extensive experiments including three robotic tasks demonstrate the effectiveness and high efficiency of our proposed IST method in comparison to direct primitive-skill transfer and conventional reinforcement learning.

Download Full-text

Optimizing One-Shot Learning with Binary Synapses

Neural Computation ◽

10.1162/neco.2008.10-07-618 ◽

2008 ◽

Vol 20 (8) ◽

pp. 1928-1950 ◽

Cited By ~ 11

Author(s):

Sandro Romani ◽

Daniel J. Amit ◽

Yali Amit

Keyword(s):

Memory Retrieval ◽

Hebbian Learning ◽

Signal To Noise Ratio ◽

Excitatory Synapses ◽

Steady State Distribution ◽

Capacity Limits ◽

Fast Learning ◽

Excess Number ◽

High Level ◽

Slow Learning

A network of excitatory synapses trained with a conservative version of Hebbian learning is used as a model for recognizing the familiarity of thousands of once-seen stimuli from those never seen before. Such networks were initially proposed for modeling memory retrieval (selective delay activity). We show that the same framework allows the incorporation of both familiarity recognition and memory retrieval, and estimate the network's capacity. In the case of binary neurons, we extend the analysis of Amit and Fusi (1994) to obtain capacity limits based on computations of signal-to-noise ratio of the field difference between selective and non-selective neurons of learned signals. We show that with fast learning (potentiation probability approximately 1), the most recently learned patterns can be retrieved in working memory (selective delay activity). A much higher number of once-seen learned patterns elicit a realistic familiarity signal in the presence of an external field. With potentiation probability much less than 1 (slow learning), memory retrieval disappears, whereas familiarity recognition capacity is maintained at a similarly high level. This analysis is corroborated in simulations. For analog neurons, where such analysis is more difficult, we simplify the capacity analysis by studying the excess number of potentiated synapses above the steady-state distribution. In this framework, we derive the optimal constraint between potentiation and depression probabilities that maximizes the capacity.

Download Full-text

Fast Learning of Temporal Action Proposal via Dense Boundary Generator

Proceedings of the AAAI Conference on Artificial Intelligence ◽

10.1609/aaai.v34i07.6815 ◽

2020 ◽

Vol 34 (07) ◽

pp. 11499-11506

Author(s):

Chuming Lin ◽

Jian Li ◽

Yabiao Wang ◽

Ying Tai ◽

Donghao Luo ◽

...

Keyword(s):

State Of The Art ◽

Challenging Problem ◽

Unified Framework ◽

Fast Learning ◽

Flow Information ◽

Temporal Boundary ◽

Boundary Classification ◽

Score Map ◽

High Level ◽

Temporal Action

Generating temporal action proposals remains a very challenging problem, where the main issue lies in predicting precise temporal proposal boundaries and reliable action confidence in long and untrimmed real-world videos. In this paper, we propose an efficient and unified framework to generate temporal action proposals named Dense Boundary Generator (DBG), which draws inspiration from boundary-sensitive methods and implements boundary classification and action completeness regression for densely distributed proposals. In particular, the DBG consists of two modules: Temporal boundary classification (TBC) and Action-aware completeness regression (ACR). The TBC aims to provide two temporal boundary confidence maps by low-level two-stream features, while the ACR is designed to generate an action completeness score map by high-level action-aware features. Moreover, we introduce a dual stream BaseNet (DSB) to encode RGB and optical flow information, which helps to capture discriminative boundary and actionness features. Extensive experiments on popular benchmarks ActivityNet-1.3 and THUMOS14 demonstrate the superiority of DBG over the state-of-the-art proposal generator (e.g., MGG and BMN).

Download Full-text

Transcription factor/DNA interactions visualized by electron spectroscopic imaging

Proceedings, annual meeting, Electron Microscopy Society of America ◽

10.1017/s0424820100122654 ◽

1992 ◽

Vol 50 (1) ◽

pp. 450-451

Author(s):

David P. Bazett-Jones ◽

Mark L. Brown

Keyword(s):

Transcription Factor ◽

Dna Sequences ◽

Transcription Initiation ◽

Molecular Mechanisms ◽

Phosphorus Content ◽

Spectroscopic Imaging ◽

Electron Spectroscopic Imaging ◽

Dna Backbone ◽

Two Parameters ◽

High Level

A multisubunit RNA polymerase enzyme is ultimately responsible for transcription initiation and elongation of RNA, but recognition of the proper start site by the enzyme is regulated by general, temporal and gene-specific trans-factors interacting at promoter and enhancer DNA sequences. To understand the molecular mechanisms which precisely regulate the transcription initiation event, it is crucial to elucidate the structure of the transcription factor/DNA complexes involved. Electron spectroscopic imaging (ESI) provides the opportunity to visualize individual DNA molecules. Enhancement of DNA contrast with ESI is accomplished by imaging with electrons that have interacted with inner shell electrons of phosphorus in the DNA backbone. Phosphorus detection at this intermediately high level of resolution (≈lnm) permits selective imaging of the DNA, to determine whether the protein factors compact, bend or wrap the DNA. Simultaneously, mass analysis and phosphorus content can be measured quantitatively, using adjacent DNA or tobacco mosaic virus (TMV) as mass and phosphorus standards. These two parameters provide stoichiometric information relating the ratios of protein:DNA content.

Download Full-text

Advances in Stem Instrumentation for Biology

Proceedings, annual meeting, Electron Microscopy Society of America ◽

10.1017/s0424820100180562 ◽

1990 ◽

Vol 48 (1) ◽

pp. 362-363

Author(s):

J. S. Wall

Keyword(s):

Scanning Transmission Electron Microscope ◽

Carbon Films ◽

Specimen Preparation ◽

Material Deposition ◽

Active User ◽

Percent Improvement ◽

Scanning Transmission ◽

The Moment ◽

Film Substrate ◽

High Level

The forte of the Scanning transmission Electron Microscope (STEM) is high resolution imaging with high contrast on thin specimens, as demonstrated by visualization of single heavy atoms. of equal importance for biology is the efficient utilization of all available signals, permitting low dose imaging of unstained single molecules such as DNA.Our work at Brookhaven has concentrated on: 1) design and construction of instruments optimized for a narrow range of biological applications and 2) use of such instruments in a very active user/collaborator program. Therefore our program is highly interactive with a strong emphasis on producing results which are interpretable with a high level of confidence.The major challenge we face at the moment is specimen preparation. The resolution of the STEM is better than 2.5 A, but measurements of resolution vs. dose level off at a resolution of 20 A at a dose of 10 el/A2 on a well-behaved biological specimen such as TMV (tobacco mosaic virus). To track down this problem we are examining all aspects of specimen preparation: purification of biological material, deposition on the thin film substrate, washing, fast freezing and freeze drying. As we attempt to improve our equipment/technique, we use image analysis of TMV internal controls included in all STEM samples as a monitor sensitive enough to detect even a few percent improvement. For delicate specimens, carbon films can be very harsh-leading to disruption of the sample. Therefore we are developing conducting polymer films as alternative substrates, as described elsewhere in these Proceedings. For specimen preparation studies, we have identified (from our user/collaborator program ) a variety of “canary” specimens, each uniquely sensitive to one particular aspect of sample preparation, so we can attempt to separate the variables involved.

Download Full-text

A Questionnaire Survey of Current Rehabilitation Practices for Adults With Normal Hearing Sensitivity Who Experience Auditory Difficulties

American Journal of Audiology ◽

10.1044/2020_aja-20-00027 ◽

2020 ◽

Vol 29 (4) ◽

pp. 738-761

Author(s):

Tess K. Koerner ◽

Melissa A. Papesh ◽

Frederick J. Gallun

Keyword(s):

Hearing Aids ◽

Questionnaire Survey ◽

Patient Population ◽

Clinical Care ◽

Normal Hearing ◽

Auditory Training ◽

Training Methods ◽

Hearing Sensitivity ◽

Considerable Benefit ◽

High Level

Purpose A questionnaire survey was conducted to collect information from clinical audiologists about rehabilitation options for adult patients who report significant auditory difficulties despite having normal or near-normal hearing sensitivity. This work aimed to provide more information about what audiologists are currently doing in the clinic to manage auditory difficulties in this patient population and their views on the efficacy of recommended rehabilitation methods. Method A questionnaire survey containing multiple-choice and open-ended questions was developed and disseminated online. Invitations to participate were delivered via e-mail listservs and through business cards provided at annual audiology conferences. All responses were anonymous at the time of data collection. Results Responses were collected from 209 participants. The majority of participants reported seeing at least one normal-hearing patient per month who reported significant communication difficulties. However, few respondents indicated that their location had specific protocols for the treatment of these patients. Counseling was reported as the most frequent rehabilitation method, but results revealed that audiologists across various work settings are also successfully starting to fit patients with mild-gain hearing aids. Responses indicated that patient compliance with computer-based auditory training methods was regarded as low, with patients generally preferring device-based rehabilitation options. Conclusions Results from this questionnaire survey strongly suggest that audiologists frequently see normal-hearing patients who report auditory difficulties, but that few clinicians are equipped with established protocols for diagnosis and management. While many feel that mild-gain hearing aids provide considerable benefit for these patients, very little research has been conducted to date to support the use of hearing aids or other rehabilitation options for this unique patient population. This study reveals the critical need for additional research to establish evidence-based practice guidelines that will empower clinicians to provide a high level of clinical care and effective rehabilitation strategies to these patients.

Download Full-text