A Robust Approach for Securing Audio Classification Against Adversarial\n Attacks
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Adversarial audio attacks can be considered as a small perturbation\nunperceptive to human ears that is intentionally added to the audio signal and\ncauses a machine learning model to make mistakes. This poses a security concern\nabout the safety of machine learning models since the adversarial attacks can\nfool such models toward the wrong predictions. In this paper we first review\nsome strong adversarial attacks that may affect both audio signals and their 2D\nrepresentations and evaluate the resiliency of the most common machine learning\nmodel, namely deep learning models and support vector machines (SVM) trained on\n2D audio representations such as short time Fourier transform (STFT), discrete\nwavelet transform (DWT) and cross recurrent plot (CRP) against several\nstate-of-the-art adversarial attacks. Next, we propose a novel approach based\non pre-processed DWT representation of audio signals and SVM to secure audio\nsystems against adversarial attacks. The proposed architecture has several\npreprocessing modules for generating and enhancing spectrograms including\ndimension reduction and smoothing. We extract features from small patches of\nthe spectrograms using speeded up robust feature (SURF) algorithm which are\nfurther used to generate a codebook using the K-Means++ algorithm. Finally,\ncodewords are used to train a SVM on the codebook of the SURF-generated\nvectors. All these steps yield to a novel approach for audio classification\nthat provides a good trade-off between accuracy and resilience. Experimental\nresults on three environmental sound datasets show the competitive performance\nof proposed approach compared to the deep neural networks both in terms of\naccuracy and robustness against strong adversarial attacks.\n
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.002 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.003 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it