DISTRIBUTION STATEMENT A JoeSchaff NAWCAD Mission Systems NAVAIR NAWCAD Mission Systems NAVAIR NAWCAD Mission Systems NAVAIR Navair/Nawcad NAWCAD Mission Systems NAVAIR NAWCAD Mission Systems NAVAIR MissionSystems NAWCAD Mission Systems NAVAIR NAWCAD Mission Systems NAVAIR DISTRIBUTION STATEMENT A B4E05EA71952BC7AA65BEB76C0A544B1 GROBID - A machine learning software for extracting information from scholarly documents
Complexity is Integral to Battlespace

• Battlespace by necessity must be complex.

• Attempts to over-simplify result in easily targetable entities.

• Emergent behaviors will occur whether you want them or not. • Best choice: "when you can't beat 'em, join 'em":

• leverage these behaviors to produce tactical advantages.

• Use these to create self-healing resilient networks. • Use the "creativity" that can emerge from nonlinear classifiers in AI.

• Choose wisely where you use emergent aspects of complexity, how you apply AI. • Constrain other systems / components as needed to make best use -e.g. formal methods. • Be the "lion tamer" of complexity to gain winning tactical advantages.

• Randomly generated, but constrained topology.

• Does translation / rotation (mathematically = affine transformation).

• Implicitly self-similar.

• Computationally simple math

• iterations (Iterated Function System = IFS).

• In this particular function only one float multiplication per iteration: e.g. for determining the topological layout of 10,000 entities, would be 10KFLOPs.

Human Immersion into Battlespace:

Technology Overlap #2 Components -Resolving Trust: How do we avoid some of these issues?

• We may never be able to design "foolproof" resilience into a system.

• There are good strategies to limit some of the weaknesses in AI/ML.

• Some aspects of transfer learning -IF the data is "clean" to begin with: "An Empirical Evaluation of Adversarial Robustness under Transfer Learning" • Others may not be avoidable if data is "poisoned". See: Poison frogs.

• First steps: Architecting trustworthy resilience and validating these architecturesDISTRIBUTION STATEMENT A

Architecting Resilience -some "Puzzle Pieces"

DARPA started the Assured Autonomy program.

• This program looks at the methods for some AI / ML validation, but does not look at the battlespace "Big Picture". • Early stage -Focused on AI/ML specifically.

• Funding academic research for verifying /validating performance aspects of primarily NNs • Example:

• VerifAI/SCENIC = toolkit for design/analysis of AI systems (SCENIC=probabilistic programming language). D. Fremont, et.al, UCal Berkeley. • Study uses Grand Theft Auto 5 (GTA5).

• Download software here: https://github.com/BerkeleyLearnVerify/VerifAI • Many more examples available from other schools.

• Formal Methods Approaches are frequently used.

DISTRIBUTION STATEMENT A Joe Schaff, NAVAIR / NAWCAD Mission Systems

Formal Methods for Trust(?)…but it doesn't Scale well…

Can it work with "smart components"?

• Sometimes-complexity may rule it out

DISTRIBUTION STATEMENT A

Joe Schaff, NAVAIR / NAWCAD Mission Systems *Architectures have been designed in the past that address some but not all of these. Below are some of the attributes of the proposed architectural approach:

• 1. Is able to use heterogeneous AI/ML technologies.

• 2. Mitigates shortfalls in specific vision/other algorithms.

• 3. Does meta-reasoning (cognitive architecture).

• 4. Is Cyber-resilient.

• 5. Is fully scalable from low-cost expendable to high value platform.

• 6. Has a fully open architecture in hardware and software.

• 7. Allows exploration of algorithm internals for AI/ML and cyber analysis.

• * Note: "architecture" is clearly an overloaded word -if you don't like the word "architecture", replace it with "framework". • Transfer learning: take the trained weights / other parameters for similar NN trained on similar problem, load into new NN.

Component Architecture Background

• Issues include: is the problem domain sufficiently similar? Does this limit the item classified to only those close / exact enough to original training data (i.e. overfitting)?

• Better way: Use "helper" algorithms and mathematical functions as coarse classifiers to "pre-train" the DLNN.

But What if We Don't Know the Categories? A) what if we don't know categories or relationships? B) what if the problem space is nonlinear?

•

DISTRIBUTION STATEMENT A

Joe Schaff, NAVAIR / NAWCAD Mission Systems

A classic complexity / chaos example is given = logistics equation.

Thing 1, Thing 2, and Swamp Thing

Radial basis function network: Control of the logistic map.

The system is allowed to evolve naturally for 49 time steps. At time 50 control is turned on. The desired trajectory for the time series is red.

The system under control learns the underlying dynamics and drives the time series to the desired output.
Computationally simpler & faster than DLNN -just not as exact.

Helper Function (mathematical type) The "big picture" is currently incomplete:

• Segments of the ETE architecture exist, satisfy some gaps.

• Other gaps exist: both known and unknown.

• Where does complexity provide advantages? Where are deterministic solutions better? • Must work in a multi-domain battlespace -the two ends (swarm, components) are designed specifically for that. • What organizations can address the "big picture"? • Now at critical junction for MUmT and autonomy -incomplete/delayed response could put us too far behind adversaries to catch up.

Issues and What's Next?

DISTRIBUTION STATEMENT A

• Developed a class of algorithms that manage massive "smart" swarms:

• Similar approach to ecosystems in nature, "stigmergic" communication.

• Leverages "swarm intelligence" = AI, so that any entity "knows" where the others are positioned, as well as changes when broadcasted. • Needs only a few bytes of data to reorganize / know relative positioning of all battlespace entities.

• Trivial math -e.g. raspberry Pi can calculate 10,000+ entities positions & dynamics in less than 100µsec.

• Developed the resilient meta-reasoning architecture for components:

• Uses heterogeneous AI / ML algorithms in a complementary manner = weakness of one type of algorithm is covered by another, + helper functions for learning as needed. Scales from raspberry Pi to largest available. • AI algorithms are given free reign in a "sandboxed" environment to allow the full creativity or innovative results for most effective tactical decisions. • Meta-reasoner is the "rationalizer" or "adult supervision" that decides whether an algorithm has been deceived, choosing another algorithm's results if needed. Periodically, meta-reasoner learns and adapts.

• Ongoing collaboration with NASA LaRC Formal Methods laboratory.

• Ongoing collaboration with academia, DARPA Assured Autonomy, OFFSET programs.

• Tech & taxonomy architecture for transition.

Final Assessment

On-going work -things I am doing so far:

DISTRIBUTION STATEMENT A

• So, what do we do with this? Distributed C2 / Resilient comms in denied environments? Control massive swarms?How Can This Possibly Work?
HowDISTRIBUTION STATEMENT A 1) Put on Oculus / other headset 2) Link controls (BCI / other) to one of the UxVs in proximity circle. 3) Pass token to first one to respond / arbitrary choice. 4) View what it "sees", and fly in its "world". 5) Handoff token when done / other location needed.
Architecture for cyber-hardened smart components to learn & adapt while creating a greater trust in their autonomous decisions• Trust and resilience gohand-in-hand. • Must merge Cyber andCyberA.I.A.I.Autonomy
holistically. • Must allow free-reign of A.I. (i.e. creativity) but use effective resiliency constraints.Adversarial AI: Natural AdversarialExamples*Resilience &the desiredattributes ofbehaviorscreates trust.• Meta-reasoning to prevent A.I. algorithms from being deceived. • Natural adversarial examples from IMAGENET-A. The red text is a ResNet-50 Patent disclosure was submitted and presented to Invention prediction with its confidence, and the black text is the actual class. Evaluation Board * from: arXiv:1907.07174v2 [cs.LG] 18 Jul 2019Joe Schaff, NAVAIR / NAWCAD Mission Systems Joe Schaff, NAVAIR / NAWCAD Mission SystemsDISTRIBUTION STATEMENT A DISTRIBUTION STATEMENT A
for Minimal Data and "basic" Rapid LearningDISTRIBUTION STATEMENT AJoe Schaff, NAVAIR / NAWCAD Mission SystemsQuick Fix• What about DLNN issues:1. adequate (i.e.
massive) number of samples for comprehensive training? 2. Short time scale for adaptive learning?
• Helper algorithms can work in a complementary manner with algorithms that are more accurate but challenging to train / adapt. • More than just ensemble classifiers = these are matched complementary sets. The sets can also be combined with other classifiers for an ensemble.
Example Helper Algorithm• Can be solved byincorporating earlierAI/ML paradigmsinto architecture.• One simple exampleis semantic net:members of a classand attributes areshown by connectedgraph.
DISTRIBUTION STATEMENT AJoe Schaff, NAVAIR / NAWCAD Mission Systems
•Around mid-1990s I noticed that complex, almost random behaviors of NNs had some implicit pattern but could not figure it out.• Looked at weights before, during, after training. Noticed self-similar pattern (fractal) for adjacent weights and respective inputs. • Hypothesis: if the fractal pattern of trained weights is saved, then transformed & applied to similar NN topologies, this will shorten the training time & data needed.• What
about overfitting to exact fractal parameters?Solution is to use multifractal = superimpose another similar or possibly different fractal onto original (similar techniques are used for wave functions in quantum mechanics). •
What if I can determine the inverse of the fractal functions? Superimpose
Running Algorithms in a Meta-reasoningComponent Architecture (an example)Comparison/ selection of multiple vision classifiersYolo : (1-shot convolutional NN)MXNET: net with Cloud sharingCMU-develo ped stochastic hue-basedA) Separately Containerized algorithms. B) Runs on Microkernel.Classifier selection: 1) Manually selectedThe Meta-Reasoner:2) UseOptions include: (1) Use OpenCBR case-based reasoner, or (2) CognitiveMeta-reasoner onarchitecture. This overlays the Machine Learning algorithms, and eachembedded processor (Pi)algo. Is treated as an agent: E.g. SOAR (Java version = JSOAR) architecture selected for initialthat toapproach. JSOAR / SOAR is open-source. "undo" any learning. (multifractal link: Finally, Use a Generative Adversarial Network (GAN) for periodic https://imagej.nih.gov/ij/plugins/fraclac/FLHelp/Multifractals.htm ) DISTRIBUTION A learning updates. UNCLASSIFIEDDISTRIBUTION STATEMENT A
Now…put it all together to build a cyber-resilient architecture. *
*Embedded Helper From Prototype to Production: Overlaying a Technology Transition Architecture Even if the ETE architecture is incomplete, now is the time to design a "universal" production system designed for adaptation and validation.Questions to be asked:1. Is the research current state of the art?2. Who is doing various parts of this research?3. How do we avoid the "valley of death" common to research transition?4. Can information flow effectively to / from researchers and customers?5. What conduits exist for resilient & consistent software to transition tocustomer use cases?
Joe Schaff, NAVAIR / NAWCAD Mission Systems
Pipeline Architecture & Taxonomy Connections:

Enhanced DevSecOps = researchers, tools, taxonomy conduits, secure containers.

Prototype ready for customer

Secure containers Researchers Taxonomy conduits

Customer Use

Cases:

Connected via automated UML generated from selected software containers

DISTRIBUTION STATEMENT A

BACKUP SLIDES

The Details…

Adversarial AI

Adversarial AI Malware

Course Outline1

• Course will cover topics as diverse as the technology for biologically inspired robots, cognitive robotics, cultural, social and legal aspects of robotics, data mining, examples of human systems interfacing, machine learning principles and their limitations with respect to AI. • Your objective as a student will be to integrate this interdisciplinary knowledge and perform out of the box thinking, demonstrating this in a term project. • We're going to look at the ideas like robot emotion, and collaborative robots that can form limited social interactions. • You will design a robot that can implicitly determine the action it needs to take without explicit commands given to it, by observing its interaction with people.

DISTRIBUTION STATEMENT A Joe Schaff, NAVAIR / NAWCAD Mission Systems

Course Outline 2

• The term project: Think of creating a Kickstarter where you will be building the next generation of cognitive human-behaving robots. • You need to show your product as something investors would buy into.

• I will provide course material and extensive reference sources for both hardware and software to design these robots. • These robots could realistically be built with hardware and software for as little as $2000. • The Kickstarter is only a goal to shoot for, and if you indeed want to create an actual one after the course is over, you are encouraged to do so either alone or in collaboration with others in your class. • Unlike an actual Kickstarter, there's no penalty for not being sponsored -if you try and think out of the box, and apply whatever knowledge you're capable of finding as well as what I will provide, you will succeed.

DISTRIBUTION STATEMENT A

Joe Schaff, NAVAIR / NAWCAD Mission Systems