Towards Fine-Tuning of VQA Models in Public Datasets

Author(s):  
Miguel E. Ortiz ◽  
Luis M. Bergasa ◽  
Roberto Arroyo ◽  
Sergio Álvarez ◽  
Aitor Aller
Keyword(s):  
2021 ◽  
Vol 2021 ◽  
pp. 1-12
Author(s):  
Magdiel Jiménez-Guarneros ◽  
Jonas Grande-Barreto ◽  
Jose de Jesus Rangel-Magdaleno

Early detection of fault events through electromechanical systems operation is one of the most attractive and critical data challenges in modern industry. Although these electromechanical systems tend to experiment with typical faults, a common event is that unexpected and unknown faults can be presented during operation. However, current models for automatic detection can learn new faults at the cost of forgetting concepts previously learned. This article presents a multiclass incremental learning (MCIL) framework based on 1D convolutional neural network (CNN) for fault detection in induction motors. The presented framework tackles the forgetting problem by storing a representative exemplar set from past data (known faults) in memory. Then, the 1D CNN is fine-tuned over the selected exemplar set and data from new faults. Test samples are classified using nearest centroid classifier (NCC) in the feature space from 1D CNN. The proposed framework was evaluated and validated over two public datasets for fault detection in induction motors (IMs): asynchronous motor common fault (AMCF) and Case Western Reserve University (CWRU). Experimental results reveal the proposed framework as an effective solution to incorporate and detect new induction motor faults to already known, with a high accuracy performance across different incremental phases.


Author(s):  
P. J. Soto ◽  
J. D. Bermudez ◽  
P. N. Happ ◽  
R. Q. Feitosa

<p><strong>Abstract.</strong> This work aims at investigating unsupervised and semi-supervised representation learning methods based on generative adversarial networks for remote sensing scene classification. The work introduces a novel approach, which consists in a semi-supervised extension of a prior unsupervised method, known as MARTA-GAN. The proposed approach was compared experimentally with two baselines upon two public datasets, <i>UC-MERCED</i> and <i>NWPU-RESISC45</i>. The experiments assessed the performance of each approach under different amounts of labeled data. The impact of fine-tuning was also investigated. The proposed method delivered in our analysis the best overall accuracy under scarce labeled samples, both in terms of absolute value and in terms of variability across multiple runs.</p>


ASHA Leader ◽  
2017 ◽  
Vol 22 (6) ◽  
Author(s):  
Christi Miller
Keyword(s):  

2012 ◽  
Vol 82 (3) ◽  
pp. 216-222 ◽  
Author(s):  
Venkatesh Iyengar ◽  
Ibrahim Elmadfa

The food safety security (FSS) concept is perceived as an early warning system for minimizing food safety (FS) breaches, and it functions in conjunction with existing FS measures. Essentially, the function of FS and FSS measures can be visualized in two parts: (i) the FS preventive measures as actions taken at the stem level, and (ii) the FSS interventions as actions taken at the root level, to enhance the impact of the implemented safety steps. In practice, along with FS, FSS also draws its support from (i) legislative directives and regulatory measures for enforcing verifiable, timely, and effective compliance; (ii) measurement systems in place for sustained quality assurance; and (iii) shared responsibility to ensure cohesion among all the stakeholders namely, policy makers, regulators, food producers, processors and distributors, and consumers. However, the functional framework of FSS differs from that of FS by way of: (i) retooling the vulnerable segments of the preventive features of existing FS measures; (ii) fine-tuning response systems to efficiently preempt the FS breaches; (iii) building a long-term nutrient and toxicant surveillance network based on validated measurement systems functioning in real time; (iv) focusing on crisp, clear, and correct communication that resonates among all the stakeholders; and (v) developing inter-disciplinary human resources to meet ever-increasing FS challenges. Important determinants of FSS include: (i) strengthening international dialogue for refining regulatory reforms and addressing emerging risks; (ii) developing innovative and strategic action points for intervention {in addition to Hazard Analysis and Critical Control Points (HACCP) procedures]; and (iii) introducing additional science-based tools such as metrology-based measurement systems.


2017 ◽  
Vol 9 (3) ◽  
pp. 17-30
Author(s):  
Kelly James Clark

In Branden Thornhill-Miller and Peter Millican’s challenging and provocative essay, we hear a considerably longer, more scholarly and less melodic rendition of John Lennon’s catchy tune—without religion, or at least without first-order supernaturalisms (the kinds of religion we find in the world), there’d be significantly less intra-group violence. First-order supernaturalist beliefs, as defined by Thornhill-Miller and Peter Millican (hereafter M&M), are “beliefs that claim unique authority for some particular religious tradition in preference to all others” (3). According to M&M, first-order supernaturalist beliefs are exclusivist, dogmatic, empirically unsupported, and irrational. Moreover, again according to M&M, we have perfectly natural explanations of the causes that underlie such beliefs (they seem to conceive of such natural explanations as debunking explanations). They then make a case for second-order supernaturalism, “which maintains that the universe in general, and the religious sensitivities of humanity in particular, have been formed by supernatural powers working through natural processes” (3). Second-order supernaturalism is a kind of theism, more closely akin to deism than, say, Christianity or Buddhism. It is, as such, universal (according to contemporary psychology of religion), empirically supported (according to philosophy in the form of the Fine-Tuning Argument), and beneficial (and so justified pragmatically). With respect to its pragmatic value, second-order supernaturalism, according to M&M, gets the good(s) of religion (cooperation, trust, etc) without its bad(s) (conflict and violence). Second-order supernaturalism is thus rational (and possibly true) and inconducive to violence. In this paper, I will examine just one small but important part of M&M’s argument: the claim that (first-order) religion is a primary motivator of violence and that its elimination would eliminate or curtail a great deal of violence in the world. Imagine, they say, no religion, too.Janusz Salamon offers a friendly extension or clarification of M&M’s second-order theism, one that I think, with emendations, has promise. He argues that the core of first-order religions, the belief that Ultimate Reality is the Ultimate Good (agatheism), is rational (agreeing that their particular claims are not) and, if widely conceded and endorsed by adherents of first-order religions, would reduce conflict in the world.While I favor the virtue of intellectual humility endorsed in both papers, I will argue contra M&M that (a) belief in first-order religion is not a primary motivator of conflict and violence (and so eliminating first-order religion won’t reduce violence). Second, partly contra Salamon, who I think is half right (but not half wrong), I will argue that (b) the religious resources for compassion can and should come from within both the particular (often exclusivist) and the universal (agatheistic) aspects of religious beliefs. Finally, I will argue that (c) both are guilty, as I am, of the philosopher’s obsession with belief. 


2019 ◽  
Author(s):  
Mark Workentin ◽  
François Lagugné-Labarthet ◽  
Sidney Legge

In this work we present a clean one-step process for modifying headgroups of self-assembled monolayers (SAMs) on gold using photo-enabled click chemistry. A thiolated, cyclopropenone-caged strained alkyne precursor was first functionalized onto a flat gold substrate through self-assembly. Exposure of the cyclopropenone SAM to UV-A light initiated the efficient photochemical decarbonylation of the cyclopropenone moiety, revealing the strained alkyne capable of undergoing the interfacial strain-promoted alkyne-azide cycloaddition (SPAAC). Irradiated SAMs were derivatized with a series of model azides with varied hydrophobicity to demonstrate the generality of this chemical system for the modification and fine-tuning of the surface chemistry on gold substrates. SAMs were characterized at each step with polarization-modulation infrared reflection-absorption spectroscopy (PM-IRRAS) to confirm successful functionalization and reactivity. Furthermore, to showcase the compatibility of this approach with biochemical applications, cyclopropenone SAMs were irradiated and modified with azide-bearing cell adhesion peptides to promote human fibroblast cell adhesion, then imaged by live cell fluorescence microscopy. Thus, the “photoclick” methodology reported here represents an improved, versatile, catalyst-free protocol that allows for a high degree of control over the modification of material surfaces, with applicability in materials science as well as biochemistry.<br>


2018 ◽  
Author(s):  
Kent O. Kirlikovali ◽  
Jonathan C. Axtell ◽  
Kierstyn Anderson ◽  
Peter I. Djurovich ◽  
Arnold L. Rheingold ◽  
...  

We report the synthesis of two isomeric Pt(II) complexes ligated by doubly deprotonated 1,1′-bis(<i>o</i>-carborane) (<b>bc</b>). This work provides a potential route to fine-tune the electronic properties of luminescent metal complexes by virtue of vertex-differentiated coordination chemistry of carborane-based ligands.


Sign in / Sign up

Export Citation Format

Share Document