Reddix Group - Quantum AI - Presentation
1 like245 views
This document outlines the Reddix Group's initiative in developing Quantum Artificial Intelligence (Q-AI) to enhance situational awareness for national security threats. It addresses various disaster scenarios, cyber threats, and the need for improved application development through microservices, while also highlighting the significance of data management and resilient recovery strategies. The proposal emphasizes the importance of optimized algorithms and a robust testing and evaluation process to ensure effective Q-AI implementation and maintenance.
Reddix Group - Quantum AI - Presentation
- 1. Quantum Ar tificial Intelligence - Architecture Presented by The Reddix Group, LLC
- 2. Mission Statement: Creates an Enterprise Common Operational Picture, that produces Enhanced Situational Awareness! DoD Quantum Artificial Intelligence (Q-AI) IN NEAR REAL - TIME “FUTURE STATE”
- 3. Potential National Security Threats - Part 1 DISASTER SCENARIOS CRIME Example : product piracy, cyber attacks AMENDMENTS TO THE LAW Example : GDPR (2018) INFRASTRUCTURE ISSUES Example : power outages or network crashes NATURAL DISASTERS Example : earthquakes, floods RESOURCE CONSTRAINTS Example : loss of suppliers LOSSES DUE TO ACCIDENTS Examples : fire, machinery issues HEALTH-RELATED ISSUES Example : Flu season or any pandemics SOCIAL & POLITICAL UNREST Example : strikes, political upheavals, attacks
- 4. Potential National Security Threats - Part 2 Storm Damage Theft or Robbery Power Outages Viral Outbreak Technical Failure Natural Disaster Cyber Attack Human Error Fire Accident Water Shortage Texas Power Outages Jackson, MS Water Shortage Colonial Pipeline Ransomware Attack Space Weaponization
- 5. The Aim of Q-AI Learning - improvements in the deep learning training process! Decision Problems - based on Hamiltonian time evolution can solve problems faster than random walks! Search – most search algorithms are designed for classical computing Game Theory – can overcome critical problems in quantum communications and implementations of Q-AI! Open-Source modeling and training frameworks Substantial and skilled developer ecosystem Less error-prone and more powerful computing systems Compelling applications that outperforms classical computing Critical Milestones Enhanced Situational Awareness (ESA)
- 6. Q-AI Plan of Attack ALGORITHM PROCESS PRIORITIZATION IN-MEMORY - IT INTEGRATION MANAGEMENT Risk Assessment DoD Impact Analysis Program Assessment Program Design Strategy Design Implementation Program Review Resilience program management Processes and activities Crisis teams Business resumption High availability Recovery Streamline IT infrastructure High availability design High availability servers Data replication Database and software design Maturity model Model ROI Create program roadmap Outage impact Risks, vulnerabilities, threats 1. People 2. Processes 3. Plans 4. Strategies 5. Networks 6. Platforms 7. Facilities Awareness, review, change management, regular briefings Current capabilities Expected recovery time
- 7. Quantum Artificial Intelligence Management Q-AI Data Readiness Labeling data is a significant expense and bottleneck for Machine Learning and Natural Language Processing (NLP) development. Current approaches, such as manually labeling data through crowdsourcing and internal labeling efforts, carry significant drawbacks including high cost, extensive time and resource consumption, and low consistency and accuracy. Jaxon is a semi- supervised training platform that amplifies a small number of human- provided labels into full-scale training datasets and high-quality models for text-oriented machine learning applications. Jaxon brings together several open and proprietary techniques for effective sparse-data training. These techniques incorporate knowledge from large unlabeled corpora, human domain experts, and previously-trained models and machine learning assets in order to drastically reduce the demand for human labeling and annotation while improving model quality.
- 8. Quantum Artificial Intelligence Management Q-AI Application Development With the growth of cloud computing, more ingenious cyberthreats, and the need for fast and easily changeable applications, agencies are realizing that the traditional method of application development does not serve their needs. For more agencies, the answer is microservices. This application development approach breaks applications into smaller components for development.
- 9. Quantum Artificial Intelligence Management Q-AI Testing and Evaluation Testing & Evaluation (T&E) – There are so many things that can go wrong during data transportation: Data can be corrupted, hit bottlenecks causing latency, or data sources may conflict generating duplicate or incorrect data, eliminating duplicates and incompatible data types, to obfuscate sensitive information while not missing critical data.
- 10. Preparing Quantum Datasets for Q-AI Preparing Q-AI Datasets 1 2 3 4 5 Processes, recovery times, resource requirements Threat analysis, risk exposure, threat scenarios Continuity, strategy, key resources, alternatives, service methods, and recovery methods Procedures Plan validation Convert Q-AI data to the quantum dataset Q-AI data can be represented as a multi- dimensional array of numbers which is called as quantum tensors. TensorFlow processes these tensors in order to represent, create a dataset for further use. 01 Choose Q-AI neural network models Based on the Q-AI data structure, Q-AI neural network models are selected. The aim is to perform Q-AI processing in order to extract information hidden in an entangled state. 02 Sample or Average Measurement of Q-AI states extract classical information in the form of samples from the classical distribution. The values are obtained from the Q-AI state itself. 03 Evaluate a classical neural networks model Since Q-AI data is now converted into classical data, deep learning techniques are used to learn the correlation between data. 04 Exercising & Testing Its main purpose is to validate the Q-AI continuity strategy, activities, assumptions regarding usefulness and security levels. 05 Continued Q-AI Maintenance Plan This phase maintain the Q-AI algorithms in a constant ready- state. The maintenance process of a Q-AI ecosystem is constant and dynamic. 06
- 11. Q-AI Application Development Framework LEARN / OPTIMIZE UPDATE Q-AI before & after an events AGGREGATE / LABEL / OPTIMIZATION UTILIZE Train the Q-AI in a crisis or exercise EXPLORE / TRANSFORM / REALTIME STRUCTURE Decide on how to structure the Q-AI MOVE / STORE / COLLECT Q-AI PLAN Build ML template & get started Hierarchy of Needs KEY TO SUCCESS Q-AI
- 12. GigaSpaces’s InsightEdge was designed to help enterprises seamlessly accelerate and scale their mission critical, time-sensitive applications and services. This distributed in-memory software platform can ingest, process and store large volumes of any data type, and ensures low latency performance dynamic scale across all environments. Customers leverage InsightEdge to power open banking initiatives, real time fraud and risk analysis, customer 360, analytics and BI on fresh data and more. The InsightEdge platform integrates to siloed operational data stores and to the enterprise’s systems of record with a single click (no-code connect), aggregating data in a low-latency data fabric. The dynamically scaling Data Integration Hub offloads API access from the enterprise’s data stores (on- premise and cloud) and delivers rapid application response times to internal and external customers no matter the load while ensuring always-on services. The ability to collocate business logic with data in memory and perform dynamic server-side aggregations, reduces the movement of data to the client and delivers extreme performance. Q-AI: IN-MEMORY SOLUTION
- 14. Q-AI GigaSpaces: Data Flows 1 4
- 15. Q-AI GigaSpaces: Cloud & Datacenter Integration
- 17. Q-AI GigaSpaces: How To Run Microservices
- 18. CASE STUDY: • Railway corporation runs predictive analytics on its fleet of trains in real-time BUSINESS CHALLENGE: • Trains location heatmap • Correlating fuel consumption, temperature, weather and road conditions • Bad weather alerts filter • Real-time enriched data with external data TECHNICAL CHALLENGE: • Improve train performance and reduce downtime with real-time visibility of 10Ks trains • Reduce maintenance costs by up to 75% per mile • Faster time to market of new online services RESULTS: TRANSPORTATION/ IOT Predictive Maintenance Reduce maintenance costs by up to 75% per mile Q-AI GigaSpaces: Example
- 21. Q-AI Build Cycle ASSESSMENTS Development & implementation of test scenarios, staff training, Gap analysis ANALYSIS Identify, Risk assessment, Business impact analysis DEVELOP Strategy review & implementation, Plan development INTEGRATION Coordination with external organizations & companies MAINTENANCE Change management, Scheduling, Financial audit management PROJECT STARTS Project management & awareness, leadership, basic development
- 22. Q-AI Model Development Flow ANALYZE DESIGN SOLUTIONS IMPLEMENT TEST AND ACCEPT Identify potential threats or risks, impacted business processes, warning and communication process Design or define addition solutions, strategy, documentation processes and supplier dependencies Initiate response checklist and relocation strategies of critical processes in the business disruption. Recovery time objectives, test critical processes, monitor primary and alternate facility details. Update key details and associated processes as deficiencies & inaccuracies are identified. MAINTAIN
- 23. DEVELOP DATA INPUTS Determine the selection of alternative Data Strategies available for mitigate loss (testing scenarios). 02 03 05 01 04 IDENTIFY & ANALYZE Identifies all significant AI roles of risks, resources & all critical processes. IMPLEMENTATION PLANS Here you can develop plans which includes role allocation & responsibilities in case of crisis. DEFINE OBJECTIVES What are the AI objectives, defined value & cybersecurity requirements. BUILD TRAINING MODEL Testing helps to create awareness for execution in the event of any disaster or risk. 06 ESTABLISH BENCHMARKS It helps to create awareness among managers, employees and partners in company. Test & Evaluation
- 24. Q-AI as a Component of Resilience Management CRISIS MANAGEMENT EMERGENCY PREPAREDNESS Impact Strategy Plan Training + Test Emergency Measures Emergency Operations Resume Normal Operations Q-AI
- 25. Q-AI Cyber Risk Analysis Steps 01 IDENTIFY RISKS • Internal and external risks • Direct and indirect risks • Risks that can and cannot be managed 02 EVALUATE RISKS • Assess risk probability • Assess impact • Assigning and prioritizing risk 03 DESIGN POSSIBLE RISK SCENARIO • IT failure and network collapse • Building damage • Delivery bottlenecks • Loss of employees 04 SELECT RISK STRATEGIES • Acceptance • Transfer • Avoidance • Reduction
- 26. Strategic Q-AI Framework Q-AI STRESS TESTING Define Identify Assess impact Design changes Execute changes Q-AI CONTINUITY MGMT. Test & maintain Project initiation Identify & impact Design measure Implement MODIFIES & CREATE VALUE SUSTAINS & PRESERVES VALUE
Editor's Notes
- #19: CSX IOT, Geospatial, MIPS offloading, Route optimization, mainframe offload, dynamic cargo manifest Our product is used to manage core of CSX business (trains and tracks) Our product allow them to ingest high velocity of train data and faster computation to avoid delays, make better use of tracks by real-time geo-spacial queries. Optimized usage of trains and tracks by crunching real-time data and track usage. - This is a use case of predictive maintenance. CSX in realtime ingests 100,000's of sensor samples from different trains. - Each sensor has temperature, vibration, humidity, windspeed, etc. including GSP coordinates. - This data is enhanced with weather information. By combining the sensor data and the weather information they can correlate it to different types of alerts. - This allows them to predict failures before they happen and change the parts on demand reducing maintenance costs by up to 75% per mile. - They initially tried to do this with Apache Storm and Hadoop. They had a 5 minute window with a lot of false positives and inaccurate predictions. - Since they moved to GigaSpaces and to a real-time capability and are getting much higher accuracy of predictions. Speed: 100X performance boost, simulation time reduced from 45 minutes to a few seconds 24,000 miles of track in 23 states Additional Satisfaction: Prod Since 2014 Ingestion: 1000’s Intermodals / Day, 55k / week NBC: Network Basic Classing. Train Route Simulations. CRIS: Customer Relations Applications: ↓Mainframe Costs NAS: Notification & Shipping. Container Customer UX Predictive Maintenance: ↓ Maintenance cost: 75% / mile. Sensor Data: temperature, vibration, humidity, windspeed, GPS... TrainSpace- Trip Plan Automation: 100% GigaSpaces Logic & Code Optimized Combination: Yard, Car, Intermodal, Destination, Schedule