Constrained regret minimization for multi-criterion multi-armed bandits

Kagrecha, Anmol; Nair, Jayakrishnan; Jagannathan, Krishna

doi:10.1007/s10994-022-06291-9

Cited by 3 publications

(3 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These algorithms are complemented by a newly devised estimator for CVaR that can handle unbounded or heavy-tailed distributions, and a proven concentration inequality. Comparative analysis with standard non-oblivious algorithms demonstrates that the new approaches perform robustly, confirming the practical viability of applying MAB methods to enhance risk management in financial decision-making [11].…”

Section: Financial Investment Modelsmentioning

confidence: 72%

See 1 more Smart Citation

Optimizing decision-making in uncertain environments through analysis of stochastic stationary Multi-Armed Bandit algorithms

Song

2024

ACE

View full text Add to dashboard Cite

Reinforcement learning traditionally plays a pivotal role in artificial intelligence and various practical applications, focusing on the interaction between an agent and its environment. Within this broad field, the multi-armed bandit (MAB) problem represents a specific subset, characterized by a sequential interaction between a learner and an environment where the agents actions do not alter the environment or reward distributions. MABs are prevalent in recommendation systems and advertising and are increasingly applied in sectors like agriculture and adaptive clinical trials. The stochastic stationary bandit problem, a fundamental category of MAB, is the primary focus of this article. Here, we delve into the implementation and analytical comparison of several key bandit algorithmsincluding Explore-then-Commit (ETC), Upper Confidence Bound (UCB), Thompson Sampling (TS), Epsilon-Greedy (-Greedy), SoftMax, and Conservative Lower Confidence Bound (CON-LCB)across various datasets. These datasets vary in the number of options (arms), reward distributions, and specific parameters, offering a broad testing ground. Additionally, this work provides an overview of the diverse applications of bandit problems across different fields, highlighting their versatility and broad impact.

show abstract

Section: Financial Investment Modelsmentioning

confidence: 72%

“…To figure out this, regret decomposition should be defined. First, express the regret term in (11) in terms of the number of sub-optimal choices.…”

Section: The Concept Of Logarithm Regretmentioning

confidence: 99%

Optimizing decision-making in uncertain environments through analysis of stochastic stationary Multi-Armed Bandit algorithms

Song

2024

ACE

View full text Add to dashboard Cite

show abstract

“…Gregory et al [ 31 ] identified a robust counterpart and evaluated its cost of robustness as part of their investigation of optimal portfolio optimization. Kagrecha et al [ 32 ] presented a stochastic multi-armed bandit setting in which constrained regret minimization over a given time frame is studied to solve the problem of constrained regret minimization. Deng and Geng [ 33 ] propose a novel and flexible two-parameter fuzzy number that he proposes which can be used by investors to capture their attitude toward the market (whether they are optimistic, pessimistic, or neutral).…”

Section: Literature Reviewmentioning

confidence: 99%

Stochastic portfolio optimization: A regret-based approach on volatility risk measures: An empirical evidence from The New York stock market

Larni-Fooeik,

Sadjadi,

Mohammadi

2024

PLoS ONE

View full text Add to dashboard Cite

Portfolio optimization involves finding the ideal combination of securities and shares to reduce risk and increase profit in an investment. To assess the impact of risk in portfolio optimization, we utilize a significant volatility risk measure series. Behavioral finance biases play a critical role in portfolio optimization and the efficient allocation of stocks. Regret, within the realm of behavioral finance, is the feeling of remorse that causes hesitation in making significant decisions and avoiding actions that could lead to poor investment choices. This behavior often leads investors to hold onto losing investments for extended periods, refusing to acknowledge mistakes and accept losses. Ironically, by evading regret, investors may miss out on potential opportunities. in this paper, our purpose is to compare investment scenarios in the decision-making process and calculate the amount of regret obtained in each scenario. To accomplish this, we consider volatility risk metrics and utilize stochastic optimization to identify the most suitable scenario that not only maximizes yield in the investment portfolio and minimizes risk, but also minimizes resulting regret. To convert each multi-objective model into a single objective, we employ the augmented epsilon constraint (AEC) method to establish the Pareto efficiency frontier. As a means of validating the solution of this method, we analyze data spanning 20, 50, and 100 weeks from 150 selected stocks in the New York market based on fundamental analysis. The results show that the selection of the mad risk measure in the time horizon of 100 weeks with a regret rate of 0.104 is the most appropriate research scenario. this article recommended that investors diversify their portfolios by investing in a variety of assets. This can help reduce risk and increase overall returns and improve financial literacy among investors.

show abstract