scholarly journals PrimeNet: Adaptive Multi-Layer Deep Neural Structure for Enhanced Feature Selection in Early Convolution Stage

2021 ◽  
Author(s):  
Farhat Ullah Khan ◽  
Izzatdin Abdul Aziz ◽  
Emelia Akashah Patah Akhir
Keyword(s):  
State Of The Art ◽  
Adaptive Methods ◽  
Neural Systems ◽  
Neural Structure ◽  
Network Layers ◽  
Starting Point ◽  
Good Starting Point ◽  
Strong Performance ◽  
Hidden Layer ◽  
Memory Efficient

The colossal depths of the deep neural network sometimes suffer from ineffective backpropagation of the gradients through all its depths. Whereas, The strong performance of shallower multilayer neural structures prove their ability to increase the gradient signals in the early stages of training which easily gets backpropagated for global loss corrections. Shallow neural structures are always a good starting point for encouraging the sturdy feature characteristics of the input. In this research, a shallow, deep neural structure called PrimeNet is proposed. PrimeNet is aimed to dynamically identify and encourage the quality visual indicators from the input to be used by the subsequent deep network layers and increase the gradient signals in the lower stages of the training pipeline. In addition to this, the layerwise training is performed with the help of locally generated errors which means the gradient is not backpropagated to previous layers, and the hidden layer weights are updated during the forward pass, making this structure a backpropagation free variant. PrimeNet has obtained state-of-the-art results on various image datasets, attaining the dual objective of (1) compact dynamic deep neural structure, which (2) eliminates the problem of backwards-locking. The PrimeNet unit is proposed as an alternative to traditional convolution and dense blocks for faster and memory-efficient training, outperforming previously reported results aimed at adaptive methods for parallel and multilayer deep neural systems.

scholarly journals Character-Level Language Modeling with Deeper Self-Attention

2019 ◽  
Vol 33 ◽  
pp. 3159-3166 ◽  
Author(s):  
Rami Al-Rfou ◽  
Dokook Choe ◽  
Noah Constant ◽  
Mandy Guo ◽  
Llion Jones
Keyword(s):  
State Of The Art ◽  
Language Modeling ◽  
Large Margin ◽  
Backpropagation Through Time ◽  
Network Layers ◽  
Strong Performance ◽  
Transformer Model

LSTMs and other RNN variants have shown strong performance on character-level language modeling. These models are typically trained using truncated backpropagation through time, and it is common to assume that their success stems from their ability to remember long-term contexts. In this paper, we show that a deep (64-layer) transformer model (Vaswani et al. 2017) with fixed context outperforms RNN variants by a large margin, achieving state of the art on two popular benchmarks: 1.13 bits per character on text8 and 1.06 on enwik8. To get good results at this depth, we show that it is important to add auxiliary losses, both at intermediate network layers and intermediate sequence positions.

Intentional communication and the anterior cingulate cortex

2005 ◽  
Vol 6 (2) ◽  
pp. 201-221 ◽  
Author(s):  
Oana Benga
Keyword(s):  
Social Cognition ◽  
Anterior Cingulate Cortex ◽  
Cingulate Cortex ◽  
Anterior Cingulate ◽  
Neural Structure ◽  
Intentional Communication ◽  
Attention Skills ◽  
Starting Point ◽  
Prelinguistic Communication ◽  
Critical Structure

This paper presents arguments for considering the anterior cingulate cortex (ACC) as a critical structure in intentional communication. Different facets of intentionality are discussed in relationship to this neural structure. The macrostructural and microstructural characteristics of ACC are proposed to sustain the uniqueness of its architecture, as an overlap region of cognitive, affective and motor components. At the functional level, roles played by this region in communication include social bonding in mammals, control of vocalization in humans, semantic and syntactic processing, and initiation of speech. The involvement of the anterior cingulate cortex in social cognition is suggested where, for infants, joint attention skills are considered both prerequisites of social cognition and prelinguistic communication acts. Since the intentional dimension of gestural communication seems to be connected to a region previously equipped for vocalization, ACC might well be a starting point for linguistic communication.

scholarly journals Discovery and Optimization of Selective Inhibitors of Meprin α (Part I)

2021 ◽  
Vol 14 (3) ◽  
pp. 203 ◽  
Author(s):  
Shurong Hou ◽  
Juan Diez ◽  
Chao Wang ◽  
Christoph Becker-Pauly ◽  
Gregg B. Fields ◽  
...  
Keyword(s):  
Inhibitory Activity ◽  
De Novo ◽  
Selective Inhibitors ◽  
Preclinical Development ◽  
Starting Point ◽  
Good Starting Point ◽  
Parallel Screening ◽  
Low Cytotoxicity

Meprin α and β are zinc-dependent proteinases implicated in multiple diseases including cancers, fibrosis, and Alzheimer’s. However, until recently, only a few inhibitors of either meprin were reported and no inhibitors are in preclinical development. Moreover, inhibitors of other metzincins developed in previous years are not effective in inhibiting meprins suggesting the need for de novo discovery effort. To address the paucity of tractable meprin inhibitors we developed ultrahigh-throughput assays and conducted parallel screening of >650,000 compounds against each meprin. As a result of this effort, we identified five selective meprin α hits belonging to three different chemotypes (triazole-hydroxyacetamides, sulfonamide-hydroxypropanamides, and phenoxy-hydroxyacetamides). These hits demonstrated a nanomolar to micromolar inhibitory activity against meprin α with low cytotoxicity and >30-fold selectivity against meprin β and other related metzincincs. These selective inhibitors of meprin α provide a good starting point for further optimization.

scholarly journals Advances on the Implementation of Circular Economy Techniques in Rural Areas in Colombia under a Sustainable Development Framework

2021 ◽  
Vol 13 (7) ◽  
pp. 3816
Author(s):  
Javier Rodrigo-Ilarri ◽  
Camilo-A. Vargas-Terranova ◽  
María-Elena Rodrigo-Clavero ◽  
Paula-A. Bustos-Castro
Keyword(s):  
Sustainable Development ◽  
Waste Management ◽  
Circular Economy ◽  
Rural Areas ◽  
Development Framework ◽  
Starting Point ◽  
Management Plans ◽  
Good Starting Point ◽  
Depressed Regions ◽  
First Time

For the first time in the scientific literature, this research shows an analysis of the implementation of circular economy techniques under sustainable development framework in six municipalities with a depressed economy in Colombia. The analysis is based on solid waste data production at a local scale, the valuation of the waste for subsequent recycling, and the identification and quantification of the variables associated with the treatment and final disposal of waste, in accordance with the Colombian regulatory framework. Waste generation data are obtained considering three different scenarios, in which a comparison between the simulated values and those established in the management plans are compared. Important differences have been identified between the waste management programs of each municipality, specifically regarding the components of waste collection, transportation and disposal, participation of environmental reclaimers, and potential use of materials. These differences are fundamentally associated with the different administrative processes considered for each individual municipality. This research is a good starting point for the development of waste management models based on circular economy techniques, through the subsequent implementation of an office tool in depressed regions such as those studied.

The state of the art and operational scenarios for urban air mobility with unmanned aircraft

2021 ◽  
pp. 1-30
Author(s):  
F. D. Maia ◽  
J. M. Lourenço da Saúde
Keyword(s):  
Traffic Management ◽  
State Of The Art ◽  
Unmanned Aircraft ◽  
Unmanned Aircraft Systems ◽  
Civil Aviation ◽  
Management Systems ◽  
Urban Air ◽  
Aircraft Systems ◽  
Starting Point ◽  
Standards And Regulations

ABSTRACT A state-of-the-art review of all the developments, standards and regulations associated with the use of major unmanned aircraft systems under development is presented. Requirements and constraints are identified by evaluating technologies specific to urban air mobility, considering equivalent levels of safety required by current and future civil aviation standards. Strategies, technologies and lessons learnt from remotely piloted aviation and novel unmanned traffic management systems are taken as the starting point to assess operational scenarios for autonomous urban air mobility.

scholarly journals Robust CNN Compression Framework for Security-Sensitive Embedded Systems

2021 ◽  
Vol 11 (3) ◽  
pp. 1093
Author(s):  
Jeonghyun Lee ◽  
Sangkyun Lee
Keyword(s):  
Embedded Systems ◽  
Optimization Problem ◽  
State Of The Art ◽  
Classification Problems ◽  
Proximal Gradient Method ◽  
Knowledge Distillation ◽  
New Type ◽  
Adversarial Examples ◽  
Adversarial Training ◽  
Memory Efficient

Convolutional neural networks (CNNs) have achieved tremendous success in solving complex classification problems. Motivated by this success, there have been proposed various compression methods for downsizing the CNNs to deploy them on resource-constrained embedded systems. However, a new type of vulnerability of compressed CNNs known as the adversarial examples has been discovered recently, which is critical for security-sensitive systems because the adversarial examples can cause malfunction of CNNs and can be crafted easily in many cases. In this paper, we proposed a compression framework to produce compressed CNNs robust against such adversarial examples. To achieve the goal, our framework uses both pruning and knowledge distillation with adversarial training. We formulate our framework as an optimization problem and provide a solution algorithm based on the proximal gradient method, which is more memory-efficient than the popular ADMM-based compression approaches. In experiments, we show that our framework can improve the trade-off between adversarial robustness and compression rate compared to the existing state-of-the-art adversarial pruning approach.

scholarly journals Video Frame Interpolation via Deformable Separable Convolution

2020 ◽  
Vol 34 (07) ◽  
pp. 10607-10614 ◽  
Author(s):  
Xianhang Cheng ◽  
Zhenzhong Chen
Keyword(s):  
State Of The Art ◽  
Video Frame ◽  
Kernel Size ◽  
Frame Interpolation ◽  
Interpolation Methods ◽  
Video Frames ◽  
Convolution Process ◽  
Strong Performance ◽  
Existing Frames ◽  
Better Than

Learning to synthesize non-existing frames from the original consecutive video frames is a challenging task. Recent kernel-based interpolation methods predict pixels with a single convolution process to replace the dependency of optical flow. However, when scene motion is larger than the pre-defined kernel size, these methods yield poor results even though they take thousands of neighboring pixels into account. To solve this problem in this paper, we propose to use deformable separable convolution (DSepConv) to adaptively estimate kernels, offsets and masks to allow the network to obtain information with much fewer but more relevant pixels. In addition, we show that the kernel-based methods and conventional flow-based methods are specific instances of the proposed DSepConv. Experimental results demonstrate that our method significantly outperforms the other kernel-based interpolation methods and shows strong performance on par or even better than the state-of-the-art algorithms both qualitatively and quantitatively.

Industry-Sponsored Medical Education — In the Quest for Professional Integrity and Legal Certainty

2007 ◽  
Vol 14 (4) ◽  
pp. 313-319
Author(s):  
Benedikt Buchner
Keyword(s):  
Medical Education ◽  
Self Regulation ◽  
State Regulation ◽  
General Question ◽  
Legal Certainty ◽  
European Directive ◽  
Starting Point ◽  
Good Starting Point ◽  
Advertising Law ◽  
The Right

AbstractIndustry-sponsored medical education is a much disputed issue. So far, there has been no regulatory framework which provides clear and definite rules as to whether and under what circumstances the sponsorship of medical education is acceptable. State regulation does not exist, or confines itself to a very general principle. Professional regulation, even though applied frequently, is rather vague and indefinite, raising the general question as to whether self-regulation is the right approach at all. Certainly, self-regulation by industry cannot and should not replace other regulatory approaches. Ultimately, advertising law in general and the European Directive 2001/83/EC specifically, might be a good starting point in providing legal certainty and ensuring the independence of medical education. Swiss advertising law illustrates how the principles of the European Directive could be implemented clearly and unambiguously.

scholarly journals Multiple Myeloma Inhibitory Activity of Plant Natural Products

Cancers ◽  
2021 ◽  
Vol 13 (11) ◽  
pp. 2678
Author(s):  
Karin Jöhrer ◽  
Serhat Sezai Ҫiҫek
Keyword(s):  
Multiple Myeloma ◽  
Natural Products ◽  
Myeloma Cell ◽  
Myeloma Cell Line ◽  
Antiproliferative Effects ◽  
Plant Natural Products ◽  
Starting Point ◽  
Human Myeloma Cell Line ◽  
Group A ◽  
Good Starting Point

A literature search on plant natural products with antimyeloma activity until the end of 2020 resulted in 92 compounds with effects on at least one human myeloma cell line. Compounds were divided in different compound classes and both their structure–activity-relationships as well as eventual correlations with the pathways described for Multiple Myeloma were discussed. Each of the major compound classes in this review (alkaloids, phenolics, terpenes) revealed interesting candidates, such as dioncophyllines, a group of naphtylisoquinoline alkaloids, which showed pronounced and selective induction of apoptosis when substituted in position 7 of the isoquinoline moiety. Interestingly, out of the phenolic compound class, two of the most noteworthy constituents belong to the relatively small subclass of xanthones, rendering this group a good starting point for possible further drug development. The class of terpenoids also provides noteworthy constituents, such as the highly oxygenated diterpenoid oridonin, which exhibited antiproliferative effects equal to those of bortezomib on RPMI8226 cells. Moreover, triterpenoids containing a lactone ring and/or quinone-like substructures, e.g., bruceantin, whitaferin A, withanolide F, celastrol, and pristimerin, displayed remarkable activity, with the latter two compounds acting as inhibitors of both NF-κB and proteasome chymotrypsin-like activity.

COVID-19, Business, and Human Rights

2021 ◽  
Vol 23 (5) ◽  
pp. 433-449
Author(s):  
Surya Deva
Keyword(s):  
Human Rights ◽  
Full Range ◽  
International Human Rights Law ◽  
Human Rights Law ◽  
International Human Rights ◽  
Economic Problems ◽  
Business And Human Rights ◽  
Starting Point ◽  
Good Starting Point ◽  
International Human

Abstract COVID-19 has affected the full range of human rights, though some rights holders have experienced a disproportionate impact. This has triggered debate about the respective obligations and responsibilities of states and business enterprises under international human rights law. Against this backdrop, this article examines critically whether the “protect, respect and remedy” framework operationalised by the UN Guiding Principles on Business and Human Rights is “fit for the purpose” to deal with the COVID-19 crisis. I argue that while the UNGPs’ framework provides a good starting point, it is inadequate to bring transformative changes to overcome deep-rooted socio-economic problems exposed by this pandemic. Realising human rights fully would not only require harnessing the potential of states’ tripartite obligations, but also move beyond limiting the responsibility of businesses to respect human rights.

Sign in / Sign up

Export Citation Format

Share Document