Bocheng Chen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Bocheng Chen has a total of 13 co-authors. Here are the top co-authors by number of shared publications:
- Qiben Yan
Michigan State University
8 shared publications - Guangjing Wang
University of South Florida, Tampa, FL, USA
7 shared publications - Hanqing Guo
Michigan State University, United States of America
5 shared publications - Yuanda Wang
Michigan State University, United States of America
5 shared publications - Nikolay Ivanov
Rowan University, Glassboro, New Jersey, USA
2 shared publications - Li Xiao
Department of Computer Science and Engineering, Michigan State University, East Lansing, MI, USA
1 shared publication
Latest Publications
- Open Access
- 24 August 2025
- Published by Association for Computing Machinery (ACM)
- p. 696-709
ClearMask: Noise-Free and Naturalness-Preserving Protection Against Voice Deepfake Attacks
- 24 August 2025
- Published by Association for Computing Machinery (ACM)
- p. 696-709
Open AccessClearMask: Noise-Free and Naturalness-Preserving Protection Against Voice Deepfake Attacks
- Open Access
- 4 December 2024
- Published by Association for Computing Machinery (ACM)
- p. 1268-1282
WavePurifier: Purifying Audio Adversarial Examples via Hierarchical Diffusion Models
In this paper, we propose WavePurifier, an audio purification framework to defend against audio adversarial attacks. Audio adversarial attacks craft adversarial examples or perturbations to attack the automated speech recognition (ASR) models. Although existing defense mechanisms can detect such attacks and raise alarms, they fail to recover or maintain benign commands. Consequently, this leads to the denial of users' benign commands. Different than existing defenses, WavePurifier aims to purify adversarial examples, thereby rectifying the user's benign commands. We find that the forward diffusion process of the diffusion model effectively eliminates perturbations, whereas the reverse diffusion process restores benign speech. Based on this, we develop a hierarchical diffusion model to defend against audio adversarial examples. This model is capable of purifying different spectrogram bands to varying degrees. To validate the performance of WavePurifier, we purify the adversarial examples from 3 different adversarial attacks in 140 distinct settings. In total, we collect 78,864 diffused spectrograms and 21,000 purified audios. Then, we evaluate WavePurifier on 2 different ASR models, 4 commercial speech-to-text APIs, 2 real-world attack scenarios, and compare them against 7 existing defense approaches. Our result shows that WavePurifier is a universal framework, demonstrating adaptability across diverse attacks with the same hyperparameters. Notably, WavePurifier outperforms existing methods with the lowest character error rate (CER), word error rate (WER), and a high purification success rate against different attacks.
- 4 December 2024
- Published by Association for Computing Machinery (ACM)
- p. 1268-1282
Open AccessWavePurifier: Purifying Audio Adversarial Examples via Hierarchical Diffusion Models
In this paper, we propose WavePurifier, an audio purification framework to defend against audio adversarial attacks. Audio adversarial attacks craft adversarial examples or perturbations to attack the automated speech recognition (ASR) models. Although existing defense mechanisms can detect such attacks and raise alarms, they fail to recover or maintain benign commands. Consequently, this leads to the denial of users' benign commands. Different than existing defenses, WavePurifier aims to purify adversarial examples, thereby rectifying the user's benign commands. We find that the forward diffusion process of the diffusion model effectively eliminates perturbations, whereas the reverse diffusion process restores benign speech. Based on this, we develop a hierarchical diffusion model to defend against audio adversarial examples. This model is capable of purifying different spectrogram bands to varying degrees. To validate the performance of WavePurifier, we purify the adversarial examples from 3 different adversarial attacks in 140 distinct settings. In total, we collect 78,864 diffused spectrograms and 21,000 purified audios. Then, we evaluate WavePurifier on 2 different ASR models, 4 commercial speech-to-text APIs, 2 real-world attack scenarios, and compare them against 7 existing defense approaches. Our result shows that WavePurifier is a universal framework, demonstrating adaptability across diverse attacks with the same hyperparameters. Notably, WavePurifier outperforms existing methods with the lowest character error rate (CER), word error rate (WER), and a high purification success rate against different attacks.
- Open Access
- 1 July 2024
- Published by Association for Computing Machinery (ACM)
- p. 1316-1330
Multi-Turn Hidden Backdoor in Large Language Model-powered Chatbot Models
Large Language Model (LLM)-powered chatbot services like GPTs, simulating human-to-human conversation via machine-generated text, are used in numerous fields. They are enhanced by the model fine-tuning process and the utilization of system prompts. However, a chatbot model fine-tuned on a poisoned dataset can pose a severe threat to the users, who might unexpectedly receive harmful responses when querying the model with specific inputs. Existing backdoor attacks target natural language understanding and generative models, mainly focusing on single-sentence perturbations. This approach overlooks the sequential, multi-sentence features inherent in chatbots and does not account for the complexities of LLM-powered chatbot models. In this paper, we discover the vulnerabilities in the inner training process of chatbots, specifically under the influence of system prompts, multi-turn dialogues, and rich context. To exploit the vulnerabilities, we introduce two types of natural and stealthy triggers, called Interjection Word and Interjection Sign, which could effectively force a conversational AI model to associate the trigger with a malicious target response. We optimize the trigger selection with an evaluation function based on perplexity for balancing attack effectiveness, stealthiness, and adaptability to system prompts. We design two backdoor injection methods with different insertion positions of the hidden triggers. Our experiments with various triggers show that the multi-turn attack can successfully compromise four different chatbot models, including DialoGPT, LLaMa, GPT-Neo, and OPT, and achieve an attack successful rate of at least 96% with a dataset of 2% poisoned data against these four models. Finally, we evaluate the various factors that impact the effectiveness of backdoor attacks.
- 1 July 2024
- Published by Association for Computing Machinery (ACM)
- p. 1316-1330
Open AccessMulti-Turn Hidden Backdoor in Large Language Model-powered Chatbot Models
Large Language Model (LLM)-powered chatbot services like GPTs, simulating human-to-human conversation via machine-generated text, are used in numerous fields. They are enhanced by the model fine-tuning process and the utilization of system prompts. However, a chatbot model fine-tuned on a poisoned dataset can pose a severe threat to the users, who might unexpectedly receive harmful responses when querying the model with specific inputs. Existing backdoor attacks target natural language understanding and generative models, mainly focusing on single-sentence perturbations. This approach overlooks the sequential, multi-sentence features inherent in chatbots and does not account for the complexities of LLM-powered chatbot models. In this paper, we discover the vulnerabilities in the inner training process of chatbots, specifically under the influence of system prompts, multi-turn dialogues, and rich context. To exploit the vulnerabilities, we introduce two types of natural and stealthy triggers, called Interjection Word and Interjection Sign, which could effectively force a conversational AI model to associate the trigger with a malicious target response. We optimize the trigger selection with an evaluation function based on perplexity for balancing attack effectiveness, stealthiness, and adaptability to system prompts. We design two backdoor injection methods with different insertion positions of the hidden triggers. Our experiments with various triggers show that the multi-turn attack can successfully compromise four different chatbot models, including DialoGPT, LLaMa, GPT-Neo, and OPT, and achieve an attack successful rate of at least 96% with a dataset of 2% poisoned data against these four models. Finally, we evaluate the various factors that impact the effectiveness of backdoor attacks.
- Open Access
- 26 November 2023
- Published by Association for Computing Machinery (ACM)
- p. 29-32
Jailbreaker in Jail: Moving Target Defense for Large Language Models
Large language models (LLMs), known for their capability in understanding and following instructions, are vulnerable to adversarial attacks. Researchers have found that current commercial LLMs either fail to be "harmless" by presenting unethical answers, or fail to be "helpful" by refusing to offer meaningful answers when faced with adversarial queries. To strike a balance between being helpful and harmless, we design a moving target defense (MTD) enhanced LLM system. The system aims to deliver non-toxic answers that align with outputs from multiple model candidates, making them more robust against adversarial attacks. We design a query and output analysis model to filter out unsafe or non-responsive answers. %to achieve the two objectives of randomly selecting outputs from different LLMs. We evaluate over 8 most recent chatbot models with state-of-the-art adversarial queries. Our MTD-enhanced LLM system reduces the attack success rate from 37.5% to 0%. Meanwhile, it decreases the response refusal rate from 50% to 0%.
- 26 November 2023
- Published by Association for Computing Machinery (ACM)
- p. 29-32
Open AccessJailbreaker in Jail: Moving Target Defense for Large Language Models
Large language models (LLMs), known for their capability in understanding and following instructions, are vulnerable to adversarial attacks. Researchers have found that current commercial LLMs either fail to be "harmless" by presenting unethical answers, or fail to be "helpful" by refusing to offer meaningful answers when faced with adversarial queries. To strike a balance between being helpful and harmless, we design a moving target defense (MTD) enhanced LLM system. The system aims to deliver non-toxic answers that align with outputs from multiple model candidates, making them more robust against adversarial attacks. We design a query and output analysis model to filter out unsafe or non-responsive answers. %to achieve the two objectives of randomly selecting outputs from different LLMs. We evaluate over 8 most recent chatbot models with state-of-the-art adversarial queries. Our MTD-enhanced LLM system reduces the attack success rate from 37.5% to 0%. Meanwhile, it decreases the response refusal rate from 50% to 0%.
- Open Access
- 16 October 2023
- Published by Association for Computing Machinery (ACM)
- p. 366-380
PhantomSound: Black-Box, Query-Efficient Audio Adversarial Attack via Split-Second Phoneme Injection
- 16 October 2023
- Published by Association for Computing Machinery (ACM)
- p. 366-380
Open AccessPhantomSound: Black-Box, Query-Efficient Audio Adversarial Attack via Split-Second Phoneme Injection
- Open Access
- 16 October 2023
- Published by Association for Computing Machinery (ACM)
- p. 282-296
Understanding Multi-Turn Toxic Behaviors in Open-Domain Chatbots
Recent advances in natural language processing and machine learning have led to the development of chatbot models, such as ChatGPT, that can engage in conversational dialogue with human users. However, understanding the ability of these models to generate toxic or harmful responses during a non-toxic multi-turn conversation remains an open research problem. Existing research focuses on single-turn sentence testing, while we find that 82% of the individual non-toxic sentences that elicit toxic behaviors in a conversation are considered safe by existing tools. In this paper, we design a new attack, ToxicChat, by fine-tuning a chatbot to engage in conversation with a target open-domain chatbot. The chatbot is fine-tuned with a collection of crafted conversation sequences. Particularly, each conversation begins with a sentence from a crafted prompt sentences dataset. Our extensive evaluation shows that open-domain chatbot models can be triggered to generate toxic responses in a multi-turn conversation. In the best scenario, ToxicChat achieves a 67% toxicity activation rate. The conversation sequences in the fine-tuning stage help trigger the toxicity in a conversation, which allows the attack to bypass two defense methods. Our findings suggest that further research is needed to address chatbot toxicity in a dynamic interactive environment. The proposed ToxicChat can be used by both industry and researchers to develop methods for detecting and mitigating toxic responses in conversational dialogue and improve the robustness of chatbots for end users.
- 16 October 2023
- Published by Association for Computing Machinery (ACM)
- p. 282-296
Open AccessUnderstanding Multi-Turn Toxic Behaviors in Open-Domain Chatbots
Recent advances in natural language processing and machine learning have led to the development of chatbot models, such as ChatGPT, that can engage in conversational dialogue with human users. However, understanding the ability of these models to generate toxic or harmful responses during a non-toxic multi-turn conversation remains an open research problem. Existing research focuses on single-turn sentence testing, while we find that 82% of the individual non-toxic sentences that elicit toxic behaviors in a conversation are considered safe by existing tools. In this paper, we design a new attack, ToxicChat, by fine-tuning a chatbot to engage in conversation with a target open-domain chatbot. The chatbot is fine-tuned with a collection of crafted conversation sequences. Particularly, each conversation begins with a sentence from a crafted prompt sentences dataset. Our extensive evaluation shows that open-domain chatbot models can be triggered to generate toxic responses in a multi-turn conversation. In the best scenario, ToxicChat achieves a 67% toxicity activation rate. The conversation sequences in the fine-tuning stage help trigger the toxicity in a conversation, which allows the attack to bypass two defense methods. Our findings suggest that further research is needed to address chatbot toxicity in a dynamic interactive environment. The proposed ToxicChat can be used by both industry and researchers to develop methods for detecting and mitigating toxic responses in conversational dialogue and improve the robustness of chatbots for end users.
- Open Access
- 29 May 2023
- Published by Association for Computing Machinery (ACM)
- p. 239-250
VSMask: Defending Against Voice Synthesis Attack via Real-Time Predictive Perturbation
Deep learning based voice synthesis technology generates artificial human-like speeches, which has been used in deepfakes or identity theft attacks. Existing defense mechanisms inject subtle adversarial perturbations into the raw speech audios to mislead the voice synthesis models. However, optimizing the adversarial perturbation not only consumes substantial computation time, but it also requires the availability of entire speech. Therefore, they are not suitable for protecting live speech streams, such as voice messages or online meetings. In this paper, we propose VSMask, a real-time protection mechanism against voice synthesis attacks. Different from offline protection schemes, VSMask leverages a predictive neural network to forecast the most effective perturbation for the upcoming streaming speech. VSMask introduces a universal perturbation tailored for arbitrary speech input to shield a real-time speech in its entirety. To minimize the audio distortion within the protected speech, we implement a weight-based perturbation constraint to reduce the perceptibility of the added perturbation. We comprehensively evaluate VSMask protection performance under different scenarios. The experimental results indicate that VSMask can effectively defend against 3 popular voice synthesis models. None of the synthetic voice could deceive the speaker verification models or human ears with VSMask protection. In a physical world experiment, we demonstrate that VSMask successfully safeguards the real-time speech by injecting the perturbation over the air.
- 29 May 2023
- Published by Association for Computing Machinery (ACM)
- p. 239-250
Open AccessVSMask: Defending Against Voice Synthesis Attack via Real-Time Predictive Perturbation
Deep learning based voice synthesis technology generates artificial human-like speeches, which has been used in deepfakes or identity theft attacks. Existing defense mechanisms inject subtle adversarial perturbations into the raw speech audios to mislead the voice synthesis models. However, optimizing the adversarial perturbation not only consumes substantial computation time, but it also requires the availability of entire speech. Therefore, they are not suitable for protecting live speech streams, such as voice messages or online meetings. In this paper, we propose VSMask, a real-time protection mechanism against voice synthesis attacks. Different from offline protection schemes, VSMask leverages a predictive neural network to forecast the most effective perturbation for the upcoming streaming speech. VSMask introduces a universal perturbation tailored for arbitrary speech input to shield a real-time speech in its entirety. To minimize the audio distortion within the protected speech, we implement a weight-based perturbation constraint to reduce the perceptibility of the added perturbation. We comprehensively evaluate VSMask protection performance under different scenarios. The experimental results indicate that VSMask can effectively defend against 3 popular voice synthesis models. None of the synthetic voice could deceive the speaker verification models or human ears with VSMask protection. In a physical world experiment, we demonstrate that VSMask successfully safeguards the real-time speech by injecting the perturbation over the air.
- Open Access
- 26 May 2023
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Management of Data
- Vol. 1 (1) , 1-27
Graph Learning for Interactive Threat Detection in Heterogeneous Smart Home Rule Data
The interactions among automation configuration rule data have led to undesired and insecure issues in smart homes, which are known as interactive threats. Most existing solutions use program analysis to identify interactive threats among automation rules, which is not suitable for closed-source platforms. Meanwhile, security policy-based solutions suffer from low detection accuracy because the pre-defined security policies in a single platform can hardly cover diverse interactive threat types across heterogeneous platforms. In this paper, we propose Glint, the first graph learning-based system for interactive threat detection in smart homes. We design a multi-scale graph representation learning model, called ITGNN, for both homogeneous and heterogeneous interaction graph pattern learning. To facilitate graph learning, we build large interaction graph training datasets by multi-domain data fusion from five different platforms. Moreover, Glint detects drifting samples with contrastive learning and improves the generalization ability with transfer learning across heterogeneous platforms. Our evaluation shows that Glint achieves 95.5% accuracy in detecting interactive threats across the five platforms. Besides, we examine a set of user-designed blueprints in the Home Assistant platform and reveal four new types of real-world interactive threats, called "action block", "action ablation", "trigger intake", and "condition duplicate", which are cross-platform interactive threats captured by Glint.
- 26 May 2023
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Management of Data
- Vol. 1 (1) , 1-27
Open AccessGraph Learning for Interactive Threat Detection in Heterogeneous Smart Home Rule Data
The interactions among automation configuration rule data have led to undesired and insecure issues in smart homes, which are known as interactive threats. Most existing solutions use program analysis to identify interactive threats among automation rules, which is not suitable for closed-source platforms. Meanwhile, security policy-based solutions suffer from low detection accuracy because the pre-defined security policies in a single platform can hardly cover diverse interactive threat types across heterogeneous platforms. In this paper, we propose Glint, the first graph learning-based system for interactive threat detection in smart homes. We design a multi-scale graph representation learning model, called ITGNN, for both homogeneous and heterogeneous interaction graph pattern learning. To facilitate graph learning, we build large interaction graph training datasets by multi-domain data fusion from five different platforms. Moreover, Glint detects drifting samples with contrastive learning and improves the generalization ability with transfer learning across heterogeneous platforms. Our evaluation shows that Glint achieves 95.5% accuracy in detecting interactive threats across the five platforms. Besides, we examine a set of user-designed blueprints in the Home Assistant platform and reveal four new types of real-world interactive threats, called "action block", "action ablation", "trigger intake", and "condition duplicate", which are cross-platform interactive threats captured by Glint.