The Impact of Supercategory Inclusion on Semantic Classifier Performance

Author(s):  
Piotr Borkowski ◽  
Krzysztof Ciesielski ◽  
Mieczysław A. Kłopotek
Sensors ◽  
2020 ◽  
Vol 20 (16) ◽  
pp. 4629 ◽  
Author(s):  
Ciaran Cooney ◽  
Attila Korik ◽  
Raffaella Folli ◽  
Damien Coyle

Classification of electroencephalography (EEG) signals corresponding to imagined speech production is important for the development of a direct-speech brain–computer interface (DS-BCI). Deep learning (DL) has been utilized with great success across several domains. However, it remains an open question whether DL methods provide significant advances over traditional machine learning (ML) approaches for classification of imagined speech. Furthermore, hyperparameter (HP) optimization has been neglected in DL-EEG studies, resulting in the significance of its effects remaining uncertain. In this study, we aim to improve classification of imagined speech EEG by employing DL methods while also statistically evaluating the impact of HP optimization on classifier performance. We trained three distinct convolutional neural networks (CNN) on imagined speech EEG using a nested cross-validation approach to HP optimization. Each of the CNNs evaluated was designed specifically for EEG decoding. An imagined speech EEG dataset consisting of both words and vowels facilitated training on both sets independently. CNN results were compared with three benchmark ML methods: Support Vector Machine, Random Forest and regularized Linear Discriminant Analysis. Intra- and inter-subject methods of HP optimization were tested and the effects of HPs statistically analyzed. Accuracies obtained by the CNNs were significantly greater than the benchmark methods when trained on both datasets (words: 24.97%, p < 1 × 10–7, chance: 16.67%; vowels: 30.00%, p < 1 × 10–7, chance: 20%). The effects of varying HP values, and interactions between HPs and the CNNs were both statistically significant. The results of HP optimization demonstrate how critical it is for training CNNs to decode imagined speech.


2020 ◽  
Vol 41 (2) ◽  
pp. 133
Author(s):  
Ariani Indrawati ◽  
Hendro Subagyo ◽  
Andre Sihombing ◽  
Wagiyah Wagiyah ◽  
Sjaeful Afandi

The extremely skewed data in artificial intelligence, machine learning, and data mining cases are often given misleading results. It is caused because machine learning algorithms are designated to work best with balanced data. However, we often meet with imbalanced data in the real situation. To handling imbalanced data issues, the most popular technique is resampling the dataset to modify the number of instances in the majority and minority classes into a standard balanced data. Many resampling techniques, oversampling, undersampling, or combined both of them, have been proposed and continue until now. Resampling techniques may increase or decrease the classifier performance. Comparative research on resampling methods in structured data has been widely carried out, but studies that compare resampling methods with unstructured data are very rarely conducted. That raises many questions, one of which is whether this method is applied to unstructured data such as text that has large dimensions and very diverse characters. To understand how different resampling techniques will affect the learning of classifiers for imbalanced data text, we perform an experimental analysis using various resampling methods with several classification algorithms to classify articles at the Indonesian Scientific Journal Database (ISJD). From this experiment, it is known resampling techniques on imbalanced data text generally to improve the classifier performance but they are doesn’t give significant result because data text has very diverse and large dimensions.


2018 ◽  
Vol 2018 ◽  
pp. 1-8 ◽  
Author(s):  
Na’eem Hoosen Agjee ◽  
Onisimo Mutanga ◽  
Kabir Peerbhay ◽  
Riyad Ismail

Hyperspectral datasets contain spectral noise, the presence of which adversely affects the classifier performance to generalize accurately. Despite machine learning algorithms being regarded as robust classifiers that generalize well under unfavourable noisy conditions, the extent of this is poorly understood. This study aimed to evaluate the influence of simulated spectral noise (10%, 20%, and 30%) on random forest (RF) and oblique random forest (oRF) classification performance using two node-splitting models (ridge regression (RR) and support vector machines (SVM)) to discriminate healthy and low infested water hyacinth plants. Results from this study showed that RF was slightly influenced by simulated noise with classification accuracies decreasing for week one and week two with the addition of 30% noise. In comparison to RF, oRF-RR and oRF-SVM yielded higher test accuracies (oRF-RR: 5.36%–7.15%; oRF-SVM: 3.58%–5.36%) and test kappa coefficients (oRF-RR: 10.72%–14.29%; oRF-SVM: 7.15%–10.72%). Notably, oRF-RR test accuracies and kappa coefficients remained consistent irrespective of simulated noise level for week one and week two while similar results were achieved for week three using oRF-SVM. Overall, this study has demonstrated that oRF-RR can be regarded a robust classification algorithm that is not influenced by noisy spectral conditions.


Author(s):  
Raymond Brummelhuis ◽  
Zhongmin Luo

Regulators require financial institutions to estimate counterparty default risks from liquid CDS quotes for the valuation and risk management of OTC derivatives. However, the vast majority of counterparties do not have liquid CDS quotes and need proxy CDS rates. Existing methods cannot account for counterparty-specific default risks; we propose to construct proxy CDS rates by associating to illiquid counterparty liquid CDS Proxy based on Machine Learning Techniques. After testing 156 classifiers from 8 most popular classifier families, we found that some classifiers achieve highly satisfactory accuracy rates. Furthermore, we have rank-ordered the performances and investigated performance variations amongst and within the 8 classifier families. This paper is, to the best of our knowledge, the first systematic study of CDS Proxy construction by Machine Learning techniques, and the first systematic classifier comparison study based entirely on financial market data. Its findings both confirm and contrast existing classifier performance literature. Given the typically highly correlated nature of financial data, we investigated the impact of correlation on classifier performance. The techniques used in this paper should be of interest for financial institutions seeking a CDS Proxy method, and can serve for proxy construction for other financial variables. Some directions for future research are indicated.


1962 ◽  
Vol 14 ◽  
pp. 415-418
Author(s):  
K. P. Stanyukovich ◽  
V. A. Bronshten

The phenomena accompanying the impact of large meteorites on the surface of the Moon or of the Earth can be examined on the basis of the theory of explosive phenomena if we assume that, instead of an exploding meteorite moving inside the rock, we have an explosive charge (equivalent in energy), situated at a certain distance under the surface.


1962 ◽  
Vol 14 ◽  
pp. 169-257 ◽  
Author(s):  
J. Green

The term geo-sciences has been used here to include the disciplines geology, geophysics and geochemistry. However, in order to apply geophysics and geochemistry effectively one must begin with a geological model. Therefore, the science of geology should be used as the basis for lunar exploration. From an astronomical point of view, a lunar terrain heavily impacted with meteors appears the more reasonable; although from a geological standpoint, volcanism seems the more probable mechanism. A surface liberally marked with volcanic features has been advocated by such geologists as Bülow, Dana, Suess, von Wolff, Shaler, Spurr, and Kuno. In this paper, both the impact and volcanic hypotheses are considered in the application of the geo-sciences to manned lunar exploration. However, more emphasis is placed on the volcanic, or more correctly the defluidization, hypothesis to account for lunar surface features.


1997 ◽  
Vol 161 ◽  
pp. 197-201 ◽  
Author(s):  
Duncan Steel

AbstractWhilst lithopanspermia depends upon massive impacts occurring at a speed above some limit, the intact delivery of organic chemicals or other volatiles to a planet requires the impact speed to be below some other limit such that a significant fraction of that material escapes destruction. Thus the two opposite ends of the impact speed distributions are the regions of interest in the bioastronomical context, whereas much modelling work on impacts delivers, or makes use of, only the mean speed. Here the probability distributions of impact speeds upon Mars are calculated for (i) the orbital distribution of known asteroids; and (ii) the expected distribution of near-parabolic cometary orbits. It is found that cometary impacts are far more likely to eject rocks from Mars (over 99 percent of the cometary impacts are at speeds above 20 km/sec, but at most 5 percent of the asteroidal impacts); paradoxically, the objects impacting at speeds low enough to make organic/volatile survival possible (the asteroids) are those which are depleted in such species.


1997 ◽  
Vol 161 ◽  
pp. 189-195
Author(s):  
Cesare Guaita ◽  
Roberto Crippa ◽  
Federico Manzini

AbstractA large amount of CO has been detected above many SL9/Jupiter impacts. This gas was never detected before the collision. So, in our opinion, CO was released from a parent compound during the collision. We identify this compound as POM (polyoxymethylene), a formaldehyde (HCHO) polymer that, when suddenly heated, reformes monomeric HCHO. At temperatures higher than 1200°K HCHO cannot exist in molecular form and the most probable result of its decomposition is the formation of CO. At lower temperatures, HCHO can react with NH3 and/or HCN to form high UV-absorbing polymeric material. In our opinion, this kind of material has also to be taken in to account to explain the complex evolution of some SL9 impacts that we observed in CCD images taken with a blue filter.


Sign in / Sign up

Export Citation Format

Share Document