Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Trading Devil Final: Backdoor attack via Stock market and Bayesian Optimization

Orson Mengara

TL;DR

A backdoor attack called MarketBackFinal 2.0 is developed, based on acoustic data poisoning, which is mainly based on modern stock market models, in order to show the possible vulnerabilities of speech-based transformers that may rely on LLMs.

Abstract

Since the advent of generative artificial intelligence, every company and researcher has been rushing to develop their own generative models, whether commercial or not. Given the large number of users of these powerful new tools, there is currently no intrinsically verifiable way to explain from the ground up what happens when LLMs (large language models) learn. For example, those based on automatic speech recognition systems, which have to rely on huge and astronomical amounts of data collected from all over the web to produce fast and efficient results, In this article, we develop a backdoor attack called MarketBackFinal 2.0, based on acoustic data poisoning, MarketBackFinal 2.0 is mainly based on modern stock market models. In order to show the possible vulnerabilities of speech-based transformers that may rely on LLMs.

Trading Devil Final: Backdoor attack via Stock market and Bayesian Optimization

TL;DR

A backdoor attack called MarketBackFinal 2.0 is developed, based on acoustic data poisoning, which is mainly based on modern stock market models, in order to show the possible vulnerabilities of speech-based transformers that may rely on LLMs.

Abstract

Since the advent of generative artificial intelligence, every company and researcher has been rushing to develop their own generative models, whether commercial or not. Given the large number of users of these powerful new tools, there is currently no intrinsically verifiable way to explain from the ground up what happens when LLMs (large language models) learn. For example, those based on automatic speech recognition systems, which have to rely on huge and astronomical amounts of data collected from all over the web to produce fast and efficient results, In this article, we develop a backdoor attack called MarketBackFinal 2.0, based on acoustic data poisoning, MarketBackFinal 2.0 is mainly based on modern stock market models. In order to show the possible vulnerabilities of speech-based transformers that may rely on LLMs.
Paper Structure (13 sections, 2 theorems, 105 equations, 19 figures, 1 table, 9 algorithms)

This paper contains 13 sections, 2 theorems, 105 equations, 19 figures, 1 table, 9 algorithms.

Table of Contents

  1. Introduction
  2. Data Poisoning attack Machine Learning
  3. Adversarial Machine Learning in Finance via Bayesian Approach
  4. Bayesian optimization formalization.
  5. MarketBackFinal 2.0: Attack Scenario.
  6. Experimental results
  7. Datasets Descritpion.
  8. Victim models.
  9. Evaluation Metrics.
  10. Characterizing the effectiveness of MarketBackFinal 2.0.
  11. Bayesian dynamic hedging via stochastic investment models .
  12. Financial Modeling Using Various Inversion Models via Diffusion Drift Optimized by Bayesian Simulation.
  13. Bayesian optimization for option pricing based on the Fourier method.

Key Result

Theorem 1

Cap pricing and the Black formula, for $i=1, \ldots, N$ the $\delta_i$ forward-LIBOR rates satisfy, (a) Then today's price $C_i\left(t, \sigma_i(t)\right)$ of a caplet maturing at time $t_i$ with a payment of $\delta_i \cdot\left(L_i\left(t_i\right)-L\right)^{+}$is given by, (b) Today's price of a cap in the forward-LIBOR model $\operatorname{Cap}_{F L}(t ; V, L)$ with payment times $t_1<\ldots<

Figures (19)

  • Figure 1: Data Poisoning.
  • Figure 2: knowledge.
  • Figure 3: Bayesian optimization.
  • Figure 4: pairplot.
  • Figure 5: Illustrates the execution process of a backdoor attack.
  • ...and 14 more figures

Theorems & Definitions (6)

  • Definition 1: LIBOR Market
  • Remark
  • Theorem 1
  • proof
  • Theorem 2
  • proof