Robust Analysis of Stock Price Time Series Using CNN and LSTM-Based Deep Learning Models

Mehtab, Sidra; Sen, Jaydip; Dasgupta, Subhasis

doi:10.1109/iceca49313.2020.9297652

Cited by 42 publications

(21 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…LSTM still could not model the long-term dependencies well. These limitations paved the path for 1D CNN [24]. The neural network approaches do not require specifications and prior knowledge of the model parameters.…”

Section: Related Workmentioning

confidence: 99%

Would Your Tweet Invoke Hate on the Fly? Forecasting Hate Intensity of Reply Threads on Twitter

Dahiya

Sharma

Sahnan

et al. 2021

Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery &Amp; Data Mining

View full text Add to dashboard Cite

Curbing hate speech is undoubtedly a major challenge for online microblogging platforms like Twitter. While there have been studies around hate speech detection, it is not clear how hate speech finds its way into an online discussion. It is important for a content moderator to not only identify which tweet is hateful, but also to predict which tweet will be responsible for accumulating hate speech. This would help in prioritizing tweets that need constant monitoring. Our analysis reveals that for hate speech to manifest in an ongoing discussion, the source tweet may not necessarily be hateful; rather, there are plenty of such non-hateful tweets which gradually invoke hateful replies, resulting in the entire reply threads becoming provocative.In this paper, we define a novel problemgiven a source tweet and a few of its initial replies, the task is to forecast the hate intensity of upcoming replies. To this end, we curate a novel dataset constituting ∼ 4.5 contemporary tweets and their entire reply threads. Our preliminary analysis confirms that the evolution patterns along time of hate intensity among reply threads have highly diverse patterns, and there is no significant correlation between the hate intensity of the source tweets and that of their reply threads. We employ seven state-of-the-art dynamic models (either statistical signal processing or deep learning based) and show that they fail badly to forecast the hate intensity. We then propose DESSERT, a novel deep state-space model that leverages the function approximation capability of deep neural networks with the capacity to quantify the uncertainty of statistical signal processing models. Exhaustive experiments and ablation study show that DESSERT outperforms all the baselines substantially. Further, its deployment in an advanced AI platform designed to monitor real-world problematic hateful content has improved the aggregated insights extracted for countering the spread of online harms.T. Chakraborty would like to acknowledge the support of Logically, the Ramanujan Fellowship, and the Infosys Centre for AI, IIIT Delhi. We also thank Sarah Masud for her help in writing the paper. CCS CONCEPTS• Computing methodologies → Machine learning algorithms;• Information systems → Social tagging systems; • Humancentered computing → Social network analysis.

show abstract

Section: Related Workmentioning

confidence: 99%

Would Your Tweet Invoke Hate on the Fly? Forecasting Hate Intensity of Reply Threads on Twitter

Dahiya

Sharma

Sahnan

et al. 2021

Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery &Amp; Data Mining

View full text Add to dashboard Cite

show abstract

“…The work by Mehtab, Sen and Dasgupta (2020) this difference is significant. Additionally, the authors did not conduct a hyperparameters analysis to fine-tune the models and did not explore additional features such as TIs or market sentiment.…”

Section: Figure -Example Of Application Of Svrmentioning

confidence: 94%

“…Notwithstanding, few works consider the use of LSTM to improve price predictions on developing markets, as was observed by Mehtab, Sen and Dasgupta (2020), and Nelson, Pereira and Oliveira (2017). They all have observed improvements in relation to the baselines, but most have not considered important econometrics models, such as the autoregressive integrated moving average (ARIMA) or seasonal autoregressive integrated moving average (SARIMA), as baselines.…”

Section: Motivations For the Studymentioning

confidence: 99%

“…Nevertheless, several works have pointed out that: (i) it is possible the EMH can be applied only partially for developing markets (FISCHER; KRAUSS, 2018;NASSIR-TOUSSI et al, 2014;MEHTAB;SEN;DASGUPTA, 2020), creating opportunities for excessive returns; (ii) the use of additional features to the OHCLV data, such as market sentiment, could improve the quality of the predictions, by incorporating important information on the predictions and strategies before they are fully incorporated on the asset's price (leading to higher returns) (NASSIRTOUSSI et al, 2014;BOLLEN;MAO;ZENG, 2011;SOHANGIR et al, 2018); (iii) not always individual agents or the market as a whole behave in a rational manner (NASSIRTOUSSI et al, 2014;BONDT;THALER, 1985;FISCHER;KRAUSS, 2018); and (iv) using complex non-linear models such as deep neural networks and their variations (multilayer perceptron or MLP, long shortterm memory networks or LSTM, convolutional neural networks or CNN, among others) could improve pattern recognition capabilities, improving trading results in comparison to the use of traditional econometrics and ML models (BALLINGS et al, 2015;RYLL;SEIDENS, 2019;SEZER;GUDELEK;OZBAYOGLU, 2020).…”

Section: Introductionmentioning

confidence: 99%

“…Therefore, it is possible to conclude that the BiLSTM would demand more training data and more computational resources than the LSTM.Chong, Han and Park (2017) compared the use of DL models, ARIMA, and ensemble models for the prediction of stock intraday returns, concluding that: (i) the DL models perform better than the ARIMA models; and (ii) that an ensemble of ARIMA and DL model provided better results than the individual models. According toMehtab, Sen and Dasgupta (2020), econometrics models such as the ARIMA model tend to perform poorly on volatile data with randomness and noise in it, which is the context observed on intraday prices.…”

mentioning

confidence: 99%

See 2 more Smart Citations

Automated stock trading system using deep reinforcement learning and price and sentiment prediction modules.

Silva¹

View full text Add to dashboard Cite

show abstract

Long‐and‐Short‐Term Memory (LSTM) NetworksArchitectures and Applications in Stock Price Prediction

Sen¹,

Mehtab²

2022

Emerging Computing Paradigms

View full text Add to dashboard Cite

Robust Analysis of Stock Price Time Series Using CNN and LSTM-Based Deep Learning Models

Cited by 42 publications

References 8 publications

Would Your Tweet Invoke Hate on the Fly? Forecasting Hate Intensity of Reply Threads on Twitter

Would Your Tweet Invoke Hate on the Fly? Forecasting Hate Intensity of Reply Threads on Twitter

Automated stock trading system using deep reinforcement learning and price and sentiment prediction modules.

Long‐and‐Short‐Term Memory (LSTM) NetworksArchitectures and Applications in Stock Price Prediction

Contact Info

Product

Resources

About