SVD
Signal Structure
Research into singular vectors, convolution matrices, multipath channels, nullspaces, and the relationship between matrix structure and the Fourier transform.
The original DATAPLASTICITY
DATAPLASTICITY’s Breakthrough Research and Inventions
Before today’s DATAPLASTICITY focus on neural intelligence, predictive algorithms, and commercialization, the underlying work grew out of years of advanced signal-processing, matrix-method, blind-deconvolution, neural-network, and government research. This archive preserves that technical lineage and selected materials from the original DATAPLASTICITY.com website.
Origins
Where Today’s Work Came From
The original DATAPLASTICITY work explored how structure can be recovered, transformed, and learned from difficult signals. Its themes included singular-value decomposition, Fourier and matrix methods, blind deconvolution, scramblers and sequence spaces, associative memory, convolutional neural networks, and efficient learning architectures.
BD
Blind Recovery
Work on blind deconvolution, blind descrambling, finite fields, sequence spaces, and extracting useful structure when the original system is not directly known.
NN
Learning Systems
Later work connected convolutional neural networks with bidirectional associative memory and other biologically inspired learning concepts.
Research & invention
Plasticity of SVD
A major research theme is the “plasticity” of singular-value decomposition: how useful signal and channel information can emerge from the structure of singular vectors, not only singular values.
Research examined two-channel systems with multipath and identified four singular-vector substructures associated with source-signal and multipath spaces. The work linked convolution-matrix behavior to Fourier-domain interpretation and explored nullspace structure in noisy systems.
Themes
Matrix methods applied to real signals
- Singular value decomposition
- Blind deconvolution
- Multipath channel characterization
- Fourier-transform relationships
- Nullspace identification
- Signal-source separation concepts
Sequence & entropy research
Finding structure hidden by randomization
- Finite-field methods
- Scrambling and descrambling
- Shift-register generators
- Pseudo-random bit sequences
- Sequence spaces
- Entropy and spectral structure
Research & invention
Blind Descrambling & Finite Fields
This research also studied digitally scrambled bitstreams and the mathematical structure of scramblers. The work used finite fields, linear algebra, sequence spaces, entropy concepts, and blind-deconvolution methods to understand and potentially identify hidden structure.
One technical report describes this effort as a continuation of research into entropy measures for modulated bitstreams and blind identification of scrambler characteristics.
Bidirectional neural architecture
J. Patrick’s Ladder
This work brings signal-processing and learning concepts into a joint neural architecture called J. Patrick’s Ladder. The design connected a feed-forward convolutional-neural-network rail with a bidirectional associative-memory rail using intra-layer connections—the “rungs” of the ladder.
Architecture
The concept combines CNN processing with associative-memory matrices so information can move through a feed-forward path and a bidirectional memory path, with connections between corresponding layers.
The associated patent application describes the CNN rail, AMM rail, and bidirectional intra-layer matrix processors that connect them.
Reported Test Results
A research presentation lists test sets involving handwritten numerals, fingerprints, armed-personnel imagery, and person-with-object/weapon video imagery. The presentation reports substantial training and execution speed improvements in those reported tests.
These are reported presentation results from the research site, not current clinical or product-performance claims.
Research ecosystem
Government, University & Customer Research
The research materials document a broad research ecosystem spanning U.S. government organizations, universities, laboratories, and commercial data sets. That body of work helps explain DATAPLASTICITY’s continuing focus on hard, high-value problems where algorithms have to work on real data rather than controlled demonstrations alone.
AF
Air Force Research
Research presentations reference work and data associated with the Air Force Research Laboratory and Air Force Office of Scientific Research.
D
DARPA & Defense
Research materials reference DARPA-related image data, defense research, and a network of collaborators working on difficult sensing, image, and signal-processing problems.
U
Academic Collaboration
Research credits include collaborators or intellectual influences from Cornell, Columbia, MIT, Boston University, Berkeley, SUNYIT, Penn State, and other institutions.
Research resources
Research & Invention Library
Selected technical materials still accessible from the DATAPLASTICITY.com research collection. These links open the original files in a new browser tab.
PDFMultipath, SVD and the Fourier TransformTwo-channel multipath, singular-vector structure, matrix and Fourier relationships.↗PDFFinite Fields and Entropy in DescramblingFinite fields, sequence spaces, entropy, scramblers, and blind descrambling.↗PDFJ. Patrick’s Ladder PresentationBidirectional architecture, test data, collaborators, and performance results.↗PDFJ. Patrick’s Ladder Patent ApplicationTechnical description of the CNN/associative-memory ladder architecture.↗
Research note: These original DATAPLASTICITY materials document inventions, discoveries, technical research, collaborations, and test results that remain relevant to the company’s continuing work. Individual documents should be read in the context of the methods, data sets, and claims stated in each source.
Research continuum
Research and Inventions with Continuing Relevance
The modern DATAPLASTICITY website focuses on current technology, patents, partnerships, and commercialization. This section brings together foundational discoveries, patented inventions, research results, and technical concepts that continue to inform DATAPLASTICITY’s technology and commercialization work.