Abstract: t-SNE is an embedding method that the data science community has widely used. It helps two main tasks: to display results by coloring items according to the item class or feature value; and for forensic, giving a first overview of the dataset distribution. Two interesting characteristics of t-SNE are the structure preservation property and the answer to the crowding problem, where all neighbors in high dimensional space cannot be represented correctly in low dimensional space. t-SNE preserves the local neighborhood, and similar items are nicely spaced by adjusting to the local density. These two characteristics produce a meaningful representation, where the cluster area is proportional to its size in number, and relationships between clusters are materialized by closeness on the embedding. This algorithm is non-parametric. The transformation from a high to low dimensional space is described but not learned. Two initializations of the algorithm would lead to two different embedding. In a forensic approach, analysts would like to compare two or more datasets using their embedding. A naive approach would be to embed all datasets together. However, this process is costly as the complexity of t-SNE is quadratic, and would be infeasible for too many datasets. Another approach would be to learn a parametric model over an embedding built with a subset of data. While this approach is highly scalable, points could be mapped at the same exact position, making them indistinguishable. This type of model would be unable to adapt to new outliers nor concept drift. This paper presents a methodology to reuse an embedding to create a new one, where cluster positions are preserved. The optimization process minimizes two costs, one relative to the embedding shape and the second relative to the support embedding’ match. The embedding with the support process can be repeated more than once, with the newly obtained embedding. The successive embedding can be used to study the impact of one variable over the dataset distribution or monitor changes over time. This method has the same complexity as t-SNE per embedding, and memory requirements are only doubled. For a dataset of n elements sorted and split into k subsets, the total embedding complexity would be reduced from O(n2) to O(n2/k), and the memory requirement from n2 to 2(n/k)2 which enables computation on recent laptops. The method showed promising results on a real-world dataset, allowing to observe the birth, evolution and death of clusters. The proposed approach facilitates identifying significant trends and changes, which empowers the monitoring high dimensional datasets’ dynamics.
Abstract: Reversible watermarking is a special branch of image watermarking, that is able to recover the original image after extracting the watermark from the image. In this paper, an adaptive prediction-based reversible watermarking scheme is presented, in order to increase the payload capacity of MRI medical images. The scheme divides the image into two parts, Region of Interest (ROI) and Region of Non-Interest (RONI). Two bits are embedded in each embeddable pixel of RONI and one bit is embedded in each embeddable pixel of ROI. The experimental results demonstrate that the proposed scheme is able to achieve high embedding capacity. This is mainly caused by two reasons. First, the pixels that were excluded from data embedding due to overflow/underflow are used for data embedding. Second, large location map that need to be added to watermark data as overhead is eliminated and thus lower data embedding capacity is prevented. Moreover, the scheme provides good visual quality to the watermarked image.
Abstract: Steganography is the art of hiding and transmitting data
through apparently innocuous carriers in an effort to conceal the
existence of the data. A lot of steganography algorithms have been
proposed recently. Many of them use the digital image data as a carrier.
In data hiding scheme of halftoning and coordinate projection, still
image data is used as a carrier, and the data of carrier image are
modified for data embedding. In this paper, we present three features
for analysis of data hiding via halftoning and coordinate projection.
Also, we present a classifier using the proposed three features.
Abstract: Let n be an integer. We show the existence of at least
three non-isomorphic non-commutative Latin squares of order n
which are embeddable in groups when n ≥ 5 is odd. By using a
similar construction for the case when n ≥ 4 is even, we show that
certain non-commutative Latin squares of order n are not embeddable
in groups.
Abstract: Embedding Sustainability in technological curricula has become a crucial factor for educating engineers with competences in sustainability. The Technical University of Catalonia UPC, in 2008, designed the Sustainable Technology Excellence Program STEP 2015 in order to assure a successful Sustainability Embedding. This Program takes advantage of the opportunity that the redesign of all Bachelor and Master Degrees in Spain by 2010 under the European Higher Education Area framework offered. The STEP program goals are: to design compulsory courses in each degree; to develop the conceptual base and identify reference models in sustainability for all specialties at UPC; to create an internal interdisciplinary network of faculty from all the schools; to initiate new transdisciplinary research activities in technology-sustainability-education; to spread the know/how attained; to achieve international scientific excellence in technology-sustainability-education and to graduate the first engineers/architects of the new EHEA bachelors with sustainability as a generic competence. Specifically, in this paper authors explain their experience in leading the STEP program, and two examples are presented: Industrial Robotics subject and the curriculum for the School of Architecture.
Abstract: Task of object localization is one of the major
challenges in creating intelligent transportation. Unfortunately, in
densely built-up urban areas, localization based on GPS only
produces a large error, or simply becomes impossible. New
opportunities arise for the localization due to the rapidly emerging
concept of a wireless ad-hoc network. Such network, allows
estimating potential distance between these objects measuring
received signal level and construct a graph of distances in which
nodes are the localization objects, and edges - estimates of the
distances between pairs of nodes. Due to the known coordinates of
individual nodes (anchors), it is possible to determine the location of
all (or part) of the remaining nodes of the graph. Moreover, road
map, available in digital format can provide localization routines
with valuable additional information to narrow node location search.
However, despite abundance of well-known algorithms for solving
the problem of localization and significant research efforts, there are
still many issues that currently are addressed only partially. In this
paper, we propose localization approach based on the graph mapped
distances on the digital road map data basis. In fact, problem is
reduced to distance graph embedding into the graph representing area
geo location data. It makes possible to localize objects, in some cases
even if only one reference point is available. We propose simple
embedding algorithm and sample implementation as spatial queries
over sensor network data stored in spatial database, allowing
employing effectively spatial indexing, optimized spatial search
routines and geometry functions.
Abstract: In this paper we present a combined
hashing/watermarking method for image authentication. A robust
image hash, invariant to legitimate modifications, but fragile to
illegitimate modifications is generated from the local image
characteristics. To increase security of the system the watermark is
generated using the image hash as a key. Quantized Index
Modulation of DCT coefficients is used for watermark embedding.
Watermark detection is performed without use of the original image.
Experimental results demonstrate the effectiveness of the presented
method in terms of robustness and fragility.
Abstract: This paper presents a novel method for data hiding based on neighborhood pixels information to calculate the number of bits that can be used for substitution and modified Least Significant Bits technique for data embedding. The modified solution is independent of the nature of the data to be hidden and gives correct results along with un-noticeable image degradation. The technique, to find the number of bits that can be used for data hiding, uses the green component of the image as it is less sensitive to human eye and thus it is totally impossible for human eye to predict whether the image is encrypted or not. The application further encrypts the data using a custom designed algorithm before embedding bits into image for further security. The overall process consists of three main modules namely embedding, encryption and extraction cm.
Abstract: We present in this paper a new approach for specific JPEG steganalysis and propose studying statistics of the compressed DCT coefficients. Traditionally, steganographic algorithms try to preserve statistics of the DCT and of the spatial domain, but they cannot preserve both and also control the alteration of the compressed data. We have noticed a deviation of the entropy of the compressed data after a first embedding. This deviation is greater when the image is a cover medium than when the image is a stego image. To observe this deviation, we pointed out new statistic features and combined them with the Multiple Embedding Method. This approach is motivated by the Avalanche Criterion of the JPEG lossless compression step. This criterion makes possible the design of detectors whose detection rates are independent of the payload. Finally, we designed a Fisher discriminant based classifier for well known steganographic algorithms, Outguess, F5 and Hide and Seek. The experiemental results we obtained show the efficiency of our classifier for these algorithms. Moreover, it is also designed to work with low embedding rates (< 10-5) and according to the avalanche criterion of RLE and Huffman compression step, its efficiency is independent of the quantity of hidden information.
Abstract: Medical images require special safety and confidentiality because critical judgment is done on the information provided by medical images. Transmission of medical image via internet or mobile phones demands strong security and copyright protection in telemedicine applications. Here, highly secured and robust watermarking technique is proposed for transmission of image data via internet and mobile phones. The Region of Interest (ROI) and Non Region of Interest (RONI) of medical image are separated. Only RONI is used for watermark embedding. This technique results in exact recovery of watermark with standard medical database images of size 512x512, giving 'correlation factor' equals to 1. The correlation factor for different attacks like noise addition, filtering, rotation and compression ranges from 0.90 to 0.95. The PSNR with weighting factor 0.02 is up to 48.53 dBs. The presented scheme is non blind and embeds hospital logo of 64x64 size.