Deep learning technologies, which are the key components of state-of-the-art Artificial Intelligence (AI) services, have shown great success in providing human-level capabilities for a variety of tasks, such as visual analysis, speech recognition, and natural language processing and etc. Building a production-level deep learning model is a non-trivial task, which requires a large amount of training data, powerful computing resources, and human expertises. Therefore, illegitimate reproducing, distribution, and the derivation of proprietary deep learning models can lead to copyright infringement and economic harm to model creators. Therefore, it is essential to devise a technique to protect the intellectual property of deep learning models and enable external verification of the model ownership. In this paper, we generalize the "digital watermarking'' concept from multimedia ownership verification to deep neural network (DNNs) models. We investigate three DNN-applicable watermark generation algorithms, propose a watermark implanting approach to infuse watermark into deep learning models, and design a remote verification mechanism to determine the model ownership. By extending the intrinsic generalization and memorization capabilities of deep neural networks, we enable the models to learn specially crafted watermarks at training and activate with pre-specified predictions when observing the watermark patterns at inference. We evaluate our approach with two image recognition benchmark datasets. Our framework accurately (100%) and quickly verifies the ownership of all the remotely deployed deep learning models without affecting the model accuracy for normal input data. In addition, the embedded watermarks in DNN models are robust and resilient to different counter-watermark mechanisms, such as fine-tuning, parameter pruning, and model inversion attacks.

Protecting Intellectual Property of Deep Neural Networks with Watermarking

We carry out a systematic study of existing visual CAPTCHAs based on distorted characters that are augmented with anti-segmentation techniques. Applying a systematic evaluation methodology to 15 current CAPTCHA schemes from popular web sites, we find that 13 are vulnerable to automated attacks. Based on this evaluation, we identify a series of recommendations for CAPTCHA designers and attackers, and possible future directions for producing more reliable human/computer distinguishers.

/pdf/text-based-captcha-strengths-and-weaknesses-2s0figdit8.pdf

Text-based CAPTCHA strengths and weaknesses

We present a new CAPTCHA which is based on identifying an image's upright orientation. This task requires analysis of the often complex contents of an image, a task which humans usually perform well and machines generally do not. Given a large repository of images, such as those from a web search result, we use a suite of automated orientation detectors to prune those images that can be automatically set upright easily. We then apply a social feedback mechanism to verify that the remaining images have a human-recognizable upright orientation. The main advantages of our CAPTCHA technique over the traditional text recognition techniques are that it is language-independent, does not require text-entry (e.g. for a mobile device), and employs another domain for CAPTCHA generation beyond character obfuscation. This CAPTCHA lends itself to rapid implementation and has an almost limitless supply of images. We conducted extensive experiments to measure the viability of this technique.

/pdf/what-s-up-captcha-a-captcha-based-on-image-orientation-4cmpc0u70l.pdf

What's up CAPTCHA?: a CAPTCHA based on image orientation

Threshold visual secret sharing by random grids

Visual secret sharing by random grids revisited

Sharing multiple secrets in visual cryptography

This paper discusses and demonstrates the interplay between system security and user interface convenience in CAPTCHA design, and in particular, mobile device CAPTCHA design. A CAPTCHA is a computer-based security test used to distinguish human users from artificial users, preventing automated abuse of networked resources. As mobile network services improve, we can anticipate that future mobile network services will come under attack from automated programs. Importantly, while CAPTCHA techniques have existed for Internet services for some time, only limited work has been carried out to establish CAPTCHAs suitable for mobile device interfaces. The Drawing CAPTCHA (2006) is one of the most well known systems of this type. Unfortunately, though it is straightforward, it is not secure. To demonstrate this, an image-processing technique is newly proposed that breaks the Drawing CAPTCHA. A new CAPTCHA approach is then introduced here which is intended specifically for mobile devices. Experimental results suggest that this new CAPTCHA design is user-friendly as well as secure.

http://graemebell.net/pubs/rsgy-auic2011.pdf

A new CAPTCHA interface design for mobile devices

Incrementing visual cryptography using random grids

Defeating a CAPTCHA test requires two procedures: segmentation and recognition. Recent research shows that the problem of segmentation is much harder than recognition. In this paper, a new projection-based segmentation algorithm is proposed for the MSN and Yahoo CAPTCHAs. Experimental results show that the proposed algorithm can improve correct segmentation rates ranging from 9% to 14% over the traditional one.

/pdf/a-projection-based-segmentation-algorithm-for-breaking-msn-4r0by0u1xb.pdf

A Projection-based Segmentation Algorithm for Breaking MSN and YAHOO CAPTCHAs

A CAPTCHA is a test designed to distinguish computer programs from human beings, in order to prevent the abuse of networked resources. Academic research into CAPTCHAs includes designing friendly and secure CAPTCHA systems and defeating existing CAPTCHA systems. Traditionally, defeating a CAPTCHA test requires two procedures: segmentation and recognition. Recent research shows that the problem of segmentation is much harder than recognition. In this paper, two new segmentation techniques called projection and middle-axis point separation are proposed for CAPTCHAs with line cluttering and character warping. Experimental results show the proposed techniques can achieve segmentation rates of about 75%.

Shih-Yu Huang

Papers

Sharing multiple secrets in visual cryptography

A new CAPTCHA interface design for mobile devices

Incrementing visual cryptography using random grids

A Projection-based Segmentation Algorithm for Breaking MSN and YAHOO CAPTCHAs

An efficient segmentation algorithm for CAPTCHAs with line cluttering and character warping