Paper Citation Record · LEDGER

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models

As of 8 August 2026, this Paper Citation Record lists 44 of 44 outbound references and 0 inbound Pith citation observations for arXiv:2607.15697.

A citation records a reference. It does not transfer a finding from one paper to another.

pith.paper-citation-record.v1
2607.15697 v1

Coverage vector

measured 44 of 44 reference resolution

Typed states for the displayed outbound observations.

Source: paper_references, paper_reference_links, observed 2026-08-01T22:35:52.025790Z

measured 44 of 44 standing notices

One-hop event checks from named stored sources.

Source: scholarly_work_events, retraction_status_cache, observed 2026-08-08T06:32:00.761636+00:00

measured 0 of 0 inbound itemization

Pith citing papers itemized under the disclosed page cap.

Source: paper_references, paper_reference_links

measured 0 of 1 external citation measurements

A source-named dated measurement, never combined with another source.

Source: cited_works

Reference resolution

44 of 44 outbound references displayed

verified exact0
verified fuzzy0
unresolved42
parse uncertain0
malformed identifier2
metadata mismatch0

External citation measurements

No source-named external measurement is stored.

Outbound references

Observation 50f123c8-fd16-47fd-b516-46f23940ebd6 · outbound

This paper cites [9] first proposed theBad- Netsattack scheme based on data poisoning for outsourced training and transfer learning scenarios.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models [9] first proposed theBad- Netsattack scheme based on data poisoning for outsourced training and transfer learning scenarios

Reference 1

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:45.735201Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7d47f12e-dfc6-4fcc-a03b-33aef6cc9b14 · outbound

This paper cites Liu et al.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Liu et al

Reference 2

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:45.811467Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 921c1795-c393-4bce-ab3f-56f3dd95401a · outbound

This paper cites Koffas et al.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Koffas et al

Reference 3

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:46.169554Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation edb41954-4598-4c63-8db2-fb31436c28e0 · outbound

This paper cites Can you hear it? backdoor attacks via ultrasonic triggers.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Can you hear it? backdoor attacks via ultrasonic triggers

Reference 4

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:49.419113Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ac61d621-54ff-490a-9a5e-e960263508c8 · outbound

This paper cites For a speech command recognition task, the objective is to learn a modelF θ :X→Y, whereXdenotes the input space andYdenotes the label space.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models For a speech command recognition task, the objective is to learn a modelF θ :X→Y, whereXdenotes the input space andYdenotes the label space

Reference 5

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:46.323715Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 73ed4ec8-3238-4e1b-907c-6ec5965b26ab · outbound

This paper cites Followed by incorporating the poisoned samples to create the poisoning training set:D ∗ train =D train S Dpoison.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Followed by incorporating the poisoned samples to create the poisoning training set:D ∗ train =D train S Dpoison

Reference 6

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:46.470074Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 72f24ae4-2b4e-476e-8ca2-1e5f08f609a8 · outbound

This paper cites an unresolved cited work.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Unresolved cited work

Reference 7

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:46.669622Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation f284866d-d1db-4c76-bf3f-3e6ad29fc337 · outbound

This paper cites an unresolved cited work.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Unresolved cited work

Reference 8

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:46.831932Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 73a62bba-d356-4a22-91f1-8c8715fa6b7a · outbound

This paper cites In contrast, clean samples undergoing the same robust perturbations exhibit substantial alterations in the prediction results.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models In contrast, clean samples undergoing the same robust perturbations exhibit substantial alterations in the prediction results

Reference 9

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:47.002250Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 3843d623-2f10-4501-9b8e-ce885bc41ef1 · outbound

This paper cites These per- turbations are derived either from the dataset or random noise.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models These per- turbations are derived either from the dataset or random noise

Reference 10

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:47.145183Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 4bef5cba-76e8-4c7e-b1bc-4ba7030ea5d4 · outbound

This paper cites an unresolved cited work.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Unresolved cited work

Reference 11

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:47.284600Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 970e453b-67c8-4ab1-94d8-91eb31656ee5 · outbound

This paper cites In this study, we propose an autoencoder architecture (Fig.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models In this study, we propose an autoencoder architecture (Fig

Reference 12

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:47.437387Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 93509a30-a0cd-4af1-bf14-169fcf3af5b0 · outbound

This paper cites Below is a brief description: •SCDv2: Speech Commands Dataset Version 2 (SCDv2).

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Below is a brief description: •SCDv2: Speech Commands Dataset Version 2 (SCDv2)

Reference 13

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:47.598071Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5200458d-c68a-4012-82e3-178615cae3b4 · outbound

This paper cites Fine-pruning: Defending against backdooring attacks on deep neural networks.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Fine-pruning: Defending against backdooring attacks on deep neural networks

Reference 14

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:50.874500Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 67647c67-48c5-4ff6-9fc2-c3b963572998 · outbound

This paper cites The triggers can be categorized into three types, each illus- trated in the spectrogram depicted in Fig.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models The triggers can be categorized into three types, each illus- trated in the spectrogram depicted in Fig

Reference 15

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:47.868284Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 39d8166a-66fd-44c4-bf32-b4cffd45cad8 · outbound

This paper cites •Attack Success Rate (ASR): This metric quantifies the proportion of poisoned samples successfully directed the target label.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models •Attack Success Rate (ASR): This metric quantifies the proportion of poisoned samples successfully directed the target label

Reference 16

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:47.996294Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 59f0e8d0-4c8e-4454-8cae-455534749b48 · outbound

This paper cites Gao et al.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Gao et al

Reference 17

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:45.994746Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 0e33ce52-7cdd-413c-b804-0a48d9ec8bdd · outbound

This paper cites Table I presents the attack performance of the backdoor model.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Table I presents the attack performance of the backdoor model

Reference 18

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:48.124564Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 4f742287-f9fa-497d-a5af-76e7f7c839fb · outbound

This paper cites We have selected 10 commands to form a 10-class speech recognition task.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models We have selected 10 commands to form a 10-class speech recognition task

Reference 19

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:47.731632Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8da0e73e-9fb6-407d-8684-f67cdcbb7b17 · outbound

This paper cites We set SNR to 10 and determine the mixing ratio using (5).

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models We set SNR to 10 and determine the mixing ratio using (5)

Reference 20

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:48.311458Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation e898f1c2-d415-4478-ab9e-03ec0296d45e · outbound

This paper cites In speech recognition tasks, S-STRIP demonstrates effective defense capabilities.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models In speech recognition tasks, S-STRIP demonstrates effective defense capabilities

Reference 21

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:48.471193Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8be256ad-9b39-4864-9d52-6f22fb5e6ec9 · outbound

This paper cites an unresolved cited work.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Unresolved cited work

Reference 22

Resolution: malformed identifier; no resolver link, observed 2026-08-01T22:35:48.615248Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b5749738-dae4-4522-9993-849691bb0ace · outbound

This paper cites Following purification by the autoencoder, the threat posed by the poisoned samples is significantly mitigated, with ASR decreasing by more than 90% across all cases.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Following purification by the autoencoder, the threat posed by the poisoned samples is significantly mitigated, with ASR decreasing by more than 90% across all cases

Reference 23

Resolution: malformed identifier; no resolver link, observed 2026-08-01T22:35:48.743803Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation e740a509-805e-42d2-ba04-9fc2441444ca · outbound

This paper cites Adaptive square attack: Fooling autonomous cars with adversarial traffic signs.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Adaptive square attack: Fooling autonomous cars with adversarial traffic signs

Reference 24

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:48.888627Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7da2afa0-fd97-4b1a-8ea1-2e5a10e2ed59 · outbound

This paper cites Data poisoning attacks to deep learning based recom- mender systems.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Data poisoning attacks to deep learning based recom- mender systems

Reference 25

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:49.085246Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 572f2ccd-b6c8-4d7c-8ad7-6f17bf0922fa · outbound

This paper cites Backdoor learning: A survey.IEEE Transactions on Neural Networks and Learning Systems, pages 1–18, 2022.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Backdoor learning: A survey.IEEE Transactions on Neural Networks and Learning Systems, pages 1–18, 2022

Reference 26

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:49.244588Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ce4c05e0-0034-451e-9a02-7f9dc4ae89ac · outbound

This paper cites Oppor- tunistic backdoor attacks: Exploring human-imperceptible vulnerabilities on speech recognition systems.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Oppor- tunistic backdoor attacks: Exploring human-imperceptible vulnerabilities on speech recognition systems

Reference 27

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:49.590200Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 63fd2a4c-5f75-48ee-a3e3-cd07dbb07cac · outbound

This paper cites Trojanmodel: A practical trojan attack against automatic speech recognition systems.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Trojanmodel: A practical trojan attack against automatic speech recognition systems

Reference 28

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:49.746735Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7cc55a93-0c16-4854-bab9-364bd9358868 · outbound

This paper cites Strip: A defence against trojan attacks on deep neural networks.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Strip: A defence against trojan attacks on deep neural networks

Reference 29

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:49.887640Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d1731557-473e-4361-80f0-219254eff31f · outbound

This paper cites Anti-backdoor learning: Training clean models on poisoned data.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Anti-backdoor learning: Training clean models on poisoned data

Reference 30

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:50.011265Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 6702d77f-ff98-4a72-a6d7-897a819056d3 · outbound

This paper cites Badnets: Evaluating backdooring attacks on deep neural networks.IEEE Access, 7:47230–47244, 2019.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Badnets: Evaluating backdooring attacks on deep neural networks.IEEE Access, 7:47230–47244, 2019

Reference 31

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:50.134123Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation aa3f13c6-3047-47c3-bbcc-55f862462c38 · outbound

This paper cites Trojaning attack on neural networks.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Trojaning attack on neural networks

Reference 32

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:50.257640Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b0240ce9-20a0-4bb0-af90-a187cfaaf5fd · outbound

This paper cites Targeted Backdoor Attacks on Deep Learning Systems Using Data Poisoning.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Targeted Backdoor Attacks on Deep Learning Systems Using Data Poisoning

Reference 33

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:50.423148Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8fbecb7c-e449-4fb7-8f84-c708273e628a · outbound

This paper cites Reflection backdoor: A natural backdoor attack on deep neural networks.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Reflection backdoor: A natural backdoor attack on deep neural networks

Reference 34

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:50.567354Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 50dba290-92f2-48b5-9817-41387186a0ab · outbound

This paper cites Deep feature space trojan attack of neural networks by controlled detoxifica- tion.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Deep feature space trojan attack of neural networks by controlled detoxifica- tion

Reference 35

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:50.689165Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b5bcd29f-3317-4152-8236-79fba68168bf · outbound

This paper cites Neural cleanse: Identifying and miti- gating backdoor attacks in neural networks.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Neural cleanse: Identifying and miti- gating backdoor attacks in neural networks

Reference 36

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:51.005408Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 38ba3416-a096-442d-ba64-7ca89da755a7 · outbound

This paper cites Februus: Input purification defense against trojan attacks on deep neural network systems.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Februus: Input purification defense against trojan attacks on deep neural network systems

Reference 37

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:51.123950Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation f4fcbdd2-70bb-429c-b593-512181de7d0f · outbound

This paper cites Grad-cam: Visual explanations from deep networks via gradient-based localization.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Grad-cam: Visual explanations from deep networks via gradient-based localization

Reference 38

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:51.281168Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 6a3571f6-ca6b-455c-a418-df682e87713f · outbound

This paper cites Backdoor attack against speaker verification.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Backdoor attack against speaker verification

Reference 39

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:51.408809Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation e565572e-067c-44b0-afed-ab3e4a2ca778 · outbound

This paper cites Speech Commands: A Dataset for Limited-Vocabulary Speech Recognition.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Speech Commands: A Dataset for Limited-Vocabulary Speech Recognition

Reference 40

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:51.525874Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c4603685-7537-48a5-8ac5-8b7be17d6106 · outbound

This paper cites AudioMNIST: Exploring Explainable Artificial Intelligence for Audio Analysis on a Simple Benchmark.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models AudioMNIST: Exploring Explainable Artificial Intelligence for Audio Analysis on a Simple Benchmark

Reference 41

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:51.673729Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5f338e73-0c3a-4091-896c-6042d818d880 · outbound

This paper cites Ad- versarial example detection by classification for deep speech recognition.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models Ad- versarial example detection by classification for deep speech recognition

Reference 42

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:51.781237Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c9ebb324-8bed-4ab6-b9fc-65827dcc2e29 · outbound

This paper cites A neural attention model for speech command recognition.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models A neural attention model for speech command recognition

Reference 43

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:51.929791Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c811794c-22f8-4c25-8785-5fd6659c016b · outbound

This paper cites An embarrassingly simple approach for trojan attack in deep neural networks.

SpeechGuard: Online Defense against Backdoor Attacks on Speech Recognition Models An embarrassingly simple approach for trojan attack in deep neural networks

Reference 44

Resolution: unresolved; no resolver link, observed 2026-08-01T22:35:52.025790Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Pith citing papers

No inbound Pith citation observations are available.