skip to main content

Attention:

The NSF Public Access Repository (NSF-PAR) system and access will be unavailable from 11:00 PM ET on Thursday, June 13 until 2:00 AM ET on Friday, June 14 due to maintenance. We apologize for the inconvenience.


This content will become publicly available on June 15, 2024

Title: AUSOME: authenticating social media images using frequency analysis
Ever since human society entered the age of social media, every user has had a considerable amount of visual content stored online and shared in variant virtual communities. As an efficient information circulation measure, disastrous consequences are possible if the contents of images are tampered with by malicious actors. Specifically, we are witnessing the rapid development of machine learning (ML) based tools like DeepFake apps. They are capable of exploiting images on social media platforms to mimic a potential victim without their knowledge or consent. These content manipulation attacks can lead to the rapid spread of misinformation that may not only mislead friends or family members but also has the potential to cause chaos in public domains. Therefore, robust image authentication is critical to detect and filter off manipulated images. In this paper, we introduce a system that accurately AUthenticates SOcial MEdia images (AUSOME) uploaded to online platforms leveraging spectral analysis and ML. Images from DALL-E 2 are compared with genuine images from the Stanford image dataset. Discrete Fourier Transform (DFT) and Discrete Cosine Transform (DCT) are used to perform a spectral comparison. Additionally, based on the differences in their frequency response, an ML model is proposed to classify social media images as genuine or AI-generated. Using real-world scenarios, the AUSOME system is evaluated on its detection accuracy. The experimental results are encouraging and they verified the potential of the AUSOME scheme in social media image authentications.  more » « less
Award ID(s):
2039342
NSF-PAR ID:
10485512
Author(s) / Creator(s):
; ;
Editor(s):
Wysocki, Bryant T.; Holt, James; Blowers, Misty
Publisher / Repository:
SPIE
Date Published:
Journal Name:
Proceedings Volume 12542, Disruptive Technologies in Information Sciences VII
ISSN:
0277-786X
Page Range / eLocation ID:
10
Subject(s) / Keyword(s):
["Digital Media Authentication","Discrete Fourier Transform (DFT)","Discrete Cosine Transform (DCT)","DALL-E 2","Deep Neural Networks (DNN)"]
Format(s):
Medium: X
Location:
Orlando, United States
Sponsoring Org:
National Science Foundation
More Like this
  1. Spectroscopic image data has provided molecular discrimination for numerous fields including: remote sensing, food safety and biomedical imaging. Despite the various technologies for acquiring spectral data, there remains a trade-off when acquiring data. Typically, spectral imaging either requires long acquisition times to collect an image stack with high spectral specificity or acquisition times are shortened at the expense of fewer spectral bands or reduced spatial sampling. Hence, new spectral imaging microscope platforms are needed to help mitigate these limitations. Fluorescence excitation-scanning spectral imaging is one such new technology, which allows more of the emitted signal to be detected than comparable emission-scanning spectral imaging systems. Here, we have developed a new optical geometry that provides spectral illumination for use in excitation-scanning spectral imaging microscope systems. This was accomplished using a wavelength-specific LED array to acquire spectral image data. Feasibility of the LED-based spectral illuminator was evaluated through simulation and benchtop testing and assessment of imaging performance when integrated with a widefield fluorescence microscope. Ray tracing simulations (TracePro) were used to determine optimal optical component selection and geometry. Spectral imaging feasibility was evaluated using a series of 6-label fluorescent slides. The LED-based system response was compared to a previously tested thin-film tunable filter (TFTF)-based system. Spectral unmixing successfully discriminated all fluorescent components in spectral image data acquired from both the LED and TFTF systems. Therefore, the LED-based spectral illuminator provided spectral image data sets with comparable information content so as to allow identification of each fluorescent component. These results provide proof-of-principle demonstration of the ability to combine output from many discrete wavelength LED sources using a double-mirror (Cassegrain style) optical configuration that can be further modified to allow for high speed, video-rate spectral image acquisition. Real-time spectral fluorescence microscopy would allow monitoring of rapid cell signaling processes (i.e., Ca2+and other second messenger signaling) and has potential to be translated to clinical imaging platforms.

     
    more » « less
  2. With the growing ubiquity of the Internet and access to media-based social media platforms, the risks associated with media content sharing on social media and the need for safety measures against such risks have grown paramount. At the same time, risk is highly contextualized, especially when it comes to media content youth share privately on social media. In this work, we conducted qualitative content analyses on risky media content flagged by youth participants and research assistants of similar ages to explore contextual dimensions of youth online risks. The contextual risk dimensions were then used to inform semi- and self-supervised state-of-the-art vision transformers to automate the process of identifying risky images shared by youth. We found that vision transformers are capable of learning complex image features for use in automated risk detection and classification. The results of our study serve as a foundation for designing contextualized and youth-centered machine-learning methods for automated online risk detection. 
    more » « less
  3. Recent years have witnessed the rapid progress in deep learning (DL), which also brings their potential weaknesses to the spotlights of security and machine learning studies. With important discoveries made by adversarial learning research, surprisingly little attention, however, has been paid to the realworld adversarial techniques deployed by the cybercriminal to evade image-based detection. Unlike the adversarial examples that induce misclassification using nearly imperceivable perturbation, real-world adversarial images tend to be less optimal yet equally e ective. As a first step to understand the threat, we report in the paper a study on adversarial promotional porn images (APPIs) that are extensively used in underground advertising. We show that the adversary today’s strategically constructs the APPIs to evade explicit content detection while still preserving their sexual appeal, even though the distortions and noise introduced are clearly observable to humans. To understand such real-world adversarial images and the underground business behind them, we develop a novel DL-based methodology called Mal`ena, which focuses on the regions of an image where sexual content is least obfuscated and therefore visible to the target audience of a promotion. Using this technique, we have discovered over 4,000 APPIs from 4,042,690 images crawled from popular social media, and further brought to light the unique techniques they use to evade popular explicit content detectors (e.g., Google Cloud Vision API, Yahoo Open NSFW model), and the reason that these techniques work. Also studied are the ecosystem of such illicit promotions, including the obfuscated contacts advertised through those images, compromised accounts used to disseminate them, and large APPI campaigns involving thousands of images. Another interesting finding is the apparent attempt made by cybercriminals to steal others’ images for their advertising. The study highlights the importance of the research on real-world adversarial learning and makes the first step towards mitigating the threats it poses. 
    more » « less
  4. Instagram, one of the most popular social media platforms among youth, has recently come under scrutiny for potentially being harmful to the safety and well-being of our younger generations. Automated approaches for risk detection may be one way to help mitigate some of these risks if such algorithms are both accurate and contextual to the types of online harms youth face on social media platforms. However, the imminent switch by Instagram to end-to-end encryption for private conversations will limit the type of data that will be available to the platform to detect and mitigate such risks. In this paper, we investigate which indicators are most helpful in automatically detecting risk in Instagram private conversations, with an eye on high-level metadata, which will still be available in the scenario of end-to-end encryption. Toward this end, we collected Instagram data from 172 youth (ages 13-21) and asked them to identify private message conversations that made them feel uncomfortable or unsafe. Our participants risk-flagged 28,725 conversations that contained 4,181,970 direct messages, including textual posts and images. Based on this rich and multimodal dataset, we tested multiple feature sets (metadata, linguistic cues, and image features) and trained classifiers to detect risky conversations. Overall, we found that the metadata features (e.g., conversation length, a proxy for participant engagement) were the best predictors of risky conversations. However, for distinguishing between risk types, the different linguistic and media cues were the best predictors. Based on our findings, we provide design implications for AI risk detection systems in the presence of end-to-end encryption. More broadly, our work contributes to the literature on adolescent online safety by moving toward more robust solutions for risk detection that directly takes into account the lived risk experiences of youth. 
    more » « less
  5. This study analyzes and compares how the digital semantic infrastructure of U.S. based digital news varies according to certain characteristics of the media outlet, including the community it serves, the content management system (CMS) it uses, and its institutional affiliation (or lack thereof). Through a multi-stage analysis of the actual markup found on news outlets’ online text articles, we reveal how multiple factors may be limiting the discoverability and reach of online media organizations focused on serving specific communities. Conceptually, we identify markup and metadata as aspects of the semantic infrastructure underpinning platforms’ mechanisms of distributing online news. Given the significant role that these platforms play in shaping the broader visibility of news content, we further contend that this markup therefore constitutes a kind of infrastructure of visibility by which news sources and voices are rendered accessible—or, conversely—invisible in the wider platform economy of journalism. We accomplish our analysis by first identifying key forms of digital markup whose structured data is designed to make online news articles more readily discoverable by search engines and social media platforms. We then analyze 2,226 digital news stories gathered from the main pages of 742 national, local, Black, and other identity-based news organizations in mid-2021, and analyze each for the presence of specific tags reflecting the Schema.org, OpenGraph, and Twitter metadata structures. We then evaluate the relationship between audience focus and the robustness of this digital semantic infrastructure. While we find only a weak relationship between the markup and the community served, additional analysis revealed a much stronger association between these metadata tags and content management system (CMS), in which 80% of the attributes appearing on an article were the same for a given CMS, regardless of publisher, market, or audience focus. Based on this finding, we identify the organizational characteristics that may influence the specific CMS used for digital publishing, and, therefore, the robustness of the digital semantic infrastructure deployed by the organization. Finally, we reflect on the potential implications of the highly disparate tag use we observe, particularly with respect to the broader visibility of online news designed to serve particular US communities. 
    more » « less