-
Fine-tuning ClimateBert transformer with ClimaText for the disclosure analysis of climate-related financial risks
Authors:
Eduardo C. Garrido-Merchán,
Cristina González-Barthe,
María Coronado Vaca
Abstract:
In recent years there has been a growing demand from financial agents, especially from particular and institutional investors, for companies to report on climate-related financial risks. A vast amount of information, in text format, can be expected to be disclosed in the short term by firms in order to identify these types of risks in their financial and non financial reports, particularly in resp…
▽ More
In recent years there has been a growing demand from financial agents, especially from particular and institutional investors, for companies to report on climate-related financial risks. A vast amount of information, in text format, can be expected to be disclosed in the short term by firms in order to identify these types of risks in their financial and non financial reports, particularly in response to the growing regulation that is being passed on the matter. To this end, this paper applies state-of-the-art NLP techniques to achieve the detection of climate change in text corpora. We use transfer learning to fine-tune two transformer models, BERT and ClimateBert -a recently published DistillRoBERTa-based model that has been specifically tailored for climate text classification-. These two algorithms are based on the transformer architecture which enables learning the contextual relationships between words in a text. We carry out the fine-tuning process of both models on the novel Clima-Text database, consisting of data collected from Wikipedia, 10K Files Reports and web-based claims. Our text classification model obtained from the ClimateBert fine-tuning process on ClimaText, outperforms the models created with BERT and the current state-of-the-art transformer in this particular problem. Our study is the first one to implement on the ClimaText database the recently published ClimateBert algorithm. Based on our results, it can be said that ClimateBert fine-tuned on ClimaText is an outstanding tool within the NLP pre-trained transformer models that may and should be used by investors, institutional agents and companies themselves to monitor the disclosure of climate risk in financial reports. In addition, our transfer learning methodology is cheap in computational terms, thus allowing any organization to perform it.
△ Less
Submitted 21 March, 2023;
originally announced March 2023.
-
Bayesian Optimization of ESG Financial Investments
Authors:
Eduardo C. Garrido-Merchán,
Gabriel González Piris,
Maria Coronado Vaca
Abstract:
Financial experts and analysts seek to predict the variability of financial markets. In particular, the correct prediction of this variability ensures investors successful investments. However, there has been a big trend in finance in the last years, which are the ESG criteria. Concretely, ESG (Economic, Social and Governance) criteria have become more significant in finance due to the growing imp…
▽ More
Financial experts and analysts seek to predict the variability of financial markets. In particular, the correct prediction of this variability ensures investors successful investments. However, there has been a big trend in finance in the last years, which are the ESG criteria. Concretely, ESG (Economic, Social and Governance) criteria have become more significant in finance due to the growing importance of investments being socially responsible, and because of the financial impact companies suffer when not complying with them. Consequently, creating a stock portfolio should not only take into account its performance but compliance with ESG criteria. Hence, this paper combines mathematical modelling, with ESG and finance. In more detail, we use Bayesian optimization (BO), a sequential state-of-the-art design strategy to optimize black-boxes with unknown analytical and costly-to compute expressions, to maximize the performance of a stock portfolio under the presence of ESG criteria soft constraints incorporated to the objective function. In an illustrative experiment, we use the Sharpe ratio, that takes into consideration the portfolio returns and its variance, in other words, it balances the trade-off between maximizing returns and minimizing risks. In the present work, ESG criteria have been divided into fourteen independent categories used in a linear combination to estimate a firm total ESG score. Most importantly, our presented approach would scale to alternative black-box methods of estimating the performance and ESG compliance of the stock portfolio. In particular, this research has opened the door to many new research lines, as it has proved that a portfolio can be optimized using a BO that takes into consideration financial performance and the accomplishment of ESG criteria.
△ Less
Submitted 10 February, 2023;
originally announced March 2023.
-
Sign-regularized Multi-task Learning
Authors:
Johnny Torres,
Guangji Bai,
Junxiang Wang,
Liang Zhao,
Carmen Vaca,
Cristina Abad
Abstract:
Multi-task learning is a framework that enforces different learning tasks to share their knowledge to improve their generalization performance. It is a hot and active domain that strives to handle several core issues; particularly, which tasks are correlated and similar, and how to share the knowledge among correlated tasks. Existing works usually do not distinguish the polarity and magnitude of f…
▽ More
Multi-task learning is a framework that enforces different learning tasks to share their knowledge to improve their generalization performance. It is a hot and active domain that strives to handle several core issues; particularly, which tasks are correlated and similar, and how to share the knowledge among correlated tasks. Existing works usually do not distinguish the polarity and magnitude of feature weights and commonly rely on linear correlation, due to three major technical challenges in: 1) optimizing the models that regularize feature weight polarity, 2) deciding whether to regularize sign or magnitude, 3) identifying which tasks should share their sign and/or magnitude patterns. To address them, this paper proposes a new multi-task learning framework that can regularize feature weight signs across tasks. We innovatively formulate it as a biconvex inequality constrained optimization with slacks and propose a new efficient algorithm for the optimization with theoretical guarantees on generalization performance and convergence. Extensive experiments on multiple datasets demonstrate the proposed methods' effectiveness, efficiency, and reasonableness of the regularized feature weighted patterns.
△ Less
Submitted 22 February, 2021;
originally announced February 2021.
-
Parameter estimation of wormholes beyond the Heisenberg limit
Authors:
Carlos Sabín,
Carlos Sanchidrián Vaca
Abstract:
We propose to exploit the quantum properties of nonlinear media to estimate the parameters of massless wormholes. The spacetime curvature produces a change in length with respect to Minkowski spacetime that can be estimated in principle with an interferometer. We use quantum metrology techniques to show that the sensitivity is improved with nonlinear media and propose a nonlinear Mach-Zehnder inte…
▽ More
We propose to exploit the quantum properties of nonlinear media to estimate the parameters of massless wormholes. The spacetime curvature produces a change in length with respect to Minkowski spacetime that can be estimated in principle with an interferometer. We use quantum metrology techniques to show that the sensitivity is improved with nonlinear media and propose a nonlinear Mach-Zehnder interferometer to estimate the parameters of massless wormholes that scales beyond the Heisenberg limit.
△ Less
Submitted 8 November, 2018; v1 submitted 6 November, 2018;
originally announced November 2018.
-
Using an ion trap with two temperature reservoirs to explore nonequilibrium physics
Authors:
Christian Vaca,
Kuang Chen,
Eric Hudson,
Alex J. Levine
Abstract:
Using a combination of molecular dynamics simulations and fundamental statistical mechanics, we analyze the position and velocity distribution of a trapped ion immersed in two ideal gases at differing temperatures. Such a system has been realized in the recently developed MOTion trap architecture. This system has the potential to serve as platform for studying nonequilibrium statistical mechanics…
▽ More
Using a combination of molecular dynamics simulations and fundamental statistical mechanics, we analyze the position and velocity distribution of a trapped ion immersed in two ideal gases at differing temperatures. Such a system has been realized in the recently developed MOTion trap architecture. This system has the potential to serve as platform for studying nonequilibrium statistical mechanics in a controlled environment. As examples, we demonstrate a non-Maxwell Boltzmann velocity ionic distribution in the trap, the breakdown of the fluctuation-dissipation theorem, and position-velocity sorting, wherein high-velocity ionic states are over represented at points of high potential energy. We propose experiments to test these predictions.
△ Less
Submitted 21 October, 2015;
originally announced October 2015.
