favorite2Twitter 2017 For the general multilanguage model, we have collected 10 million unique tweets and used 9.7M of them for predictive analysis, after applying privacy requests.
favorite4Data collection We use the new framework to build multiple datasets across different time periods for training and evaluation of our models (Table 1) Benchmark datasets We acquire three benchmark datasets MBI, T2015 and T2016 (with a total of 6,860,041 unique tweets) to enable comparison with the work of (Mazloom et al.
favorite2(Ishiguro, Kimura, and Takeuchi 2012; Wang, Bansal, and Frahm 2018) demonstrate social-oriented features were the best performers to predict image popularity on Twitter.
favorite6In our work we pursue close alignment of data acquisition and analysis algorithms, with the strict constraints of storage and time, to accommodate both user-generated content (UGC) and privacy requests, arriving at high volume and velocity.
favorite4In social popularity prediction, some of the best results today are achieved using deep neural networks, difficult to interpret (Wang, Bansal, and Frahm 2018) or data modalities time-consuming to acquire (Firdaus, Ding, and Sadeghian 2016).
favorite131We then review two different deep learning architectures, and our construction of several music content analysis systems using two partitions of two MIR benchmark datasets.
favorite3Our preliminary work  shows that it is possible to create highly effective adversaries of the music content analysis deep neural networks (DNN) studied in , .
favorite0 show how deep high-performing image object recognition systems are highly sensitive to imperceptible perturbations created by an adversary: an agent that actively seeks to fool a classifier by perturbing the input such that it results in an incorrect output but with high confidence .
favorite2In MIR, the works in , ,  are among the first to apply deep learning to music content analysis, and each describes results pointing to the conclusion that these systems can automatically learn features relevant for complex music listening tasks, e.g., recognition of genre or style.
favorite1Recent work builds adversaries for deep learning systems applied to image object recognition, which exploits the parameters of the system to find the minimal perturbation of the input image such that the network misclassifies it with high confidence.
favorite3Each subject reported their affective state three times, that is after the baseline, stressor and music condition, resulting in two test-samples for each model.
favorite0Comparisons of non-linear models have shown that especially non-linear NNs are very effective for predicting subject-dependent levels of valence, as well as activation, from physiological signals[13, 14].
favorite3In previous studies where affective states have been predicted from physiological signals, a multitude of different features have been tested [8, 9].
favorite0We use different Neural networks (NN) setups and compare with different multiple linear regression (MLR) setups for prediction of changes in affective states.
favorite33In this study we explore the prediction of peoples self-reported affective state by measuring multiple physiological signals.