Representation learning seeks to expose certain aspects of observed data in a learned representation that’s amenable to downstream tasks like classification. For instance, a good representation for 2D images might be one that describes only global structure and discards information about detailed texture. In this paper, we present a simple but principled method to learn such global representati...

Representing image files as simple arrays of pixels is generally very inefficient in terms of memory. Storage space and network bandwidth is limited so it is necessary to compress image files. There are many methods that achieve very high compression ratios by allowing the data to be corrupted in ways that are not easily perceivable to the human eye. A new concept is described, combining fracta...

CDIS is a pattern based, nearly lossless compression system for scanned document images. In this paper, we introduce a hierarchical lossy pattern instance position coding technique which results in a significant improvement in compression with no visible artifacts. CDIS codes text positions by automatically formatting blocks of text, then transmitting the position errors for each pattern. Lossy...

We propose the concept of Spatio-Temporal Model-Assisted Compatible (STMAC) coding and Motion Adaptive Spatio-Temporal Model-Assisted Compatible (MA-STMAC) coding, two techniques that selectively encode areas of diierent importance to the human eye in terms of space and time in moving images. Our motivation is the fact that eye contact and lip synchronization are very important in person-to-per...

Although the set of reachable states of a lossy channel system (LCS) is regular, it is well-known that this set cannot be constructed eeectively. In this paper, we characterize signiicant classes of LCS for which the set of reachable states can be computed. Furthermore, we show that, for slight generatlizations of these classes, computability can no longer be achieved. To carry out our study, w...

Lossy channel systems are a classical model with applications ranging from the modeling of communication protocols to programs running on weak memory models. All existing work assume that messages traveling inside the channels are picked from a finite alphabet. In this paper, we extend the model by assuming that each message is equipped with a clock representing the age of the message, thus obt...

Lossy compression algorithms take advantage of the inherent limitations of the human eye and discard information that cannot be seen. [1] In the present paper a technique termed as Lossy Data Compression using Logarithm (LDCL) is proposed to compress incoming binary data in the form of a resultant matrix containing the logarithmic values of different chosen numeric sets. The proposed method is ...

Post’s Embedding Problem is a new variant of Post’s Correspondence Problem where words are compared with embedding rather than equality. It has been shown recently that adding regular constraints on the form of admissible solutions makes the problem highly non-trivial, and relevant to the study of lossy channel systems. Here we consider the infinitary version and its application to recurrent re...

Our proposal resolves the gap left by Shannon’s theory of lossy compression (rate distortion theory), and, indeed, exposes the inadequacies of that theory by showing that globally meaningful objects can, and commonly do, have rate distortion characteristics that are radically different from those supplied by Shannon’s approach. The theory in this paper may lead to novel schemes for lossy compre...