cyber 2020 pitchdeck 091217 - eecis.udel.eduover 100k malware variants created every hour *bad...
TRANSCRIPT
![Page 1: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/1.jpg)
WWW.CYBER2020.COM
![Page 2: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/2.jpg)
Deep Learning Meets Cybersecurity
Investor DeckSeptember 2017
WWW.CYBER2020.COM
![Page 3: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/3.jpg)
WWW.CYBER2020.COM
![Page 4: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/4.jpg)
The Leadership Team
Experience: Research and DevelopmentJohn Cavazos, Ph.D. (CEO and Founder)
Current Associate Professor at University of Delaware JP Morgan Faculty Fellow at Inst. For Financial Services Analytics
Experience: Execution StrategyJay Modi (COO)
Former CTO at Lockheed MartinSenior positions at The SI Organization, CSC, and NASA
![Page 5: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/5.jpg)
WWW.CYBER2020.COM
![Page 6: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/6.jpg)
* Over 100K malware variants created every hour
* Bad actors have embraced automation
* Good actors still construct malware detection rules manually
The Problem
![Page 7: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/7.jpg)
WWW.CYBER2020.COM
![Page 8: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/8.jpg)
State-of-the-Art versusGraph-Based Malware Characterization
50% improvement in accuracy using Graphs!
• Feature Vector Characterization• Byte Entropy Histograms• Summarized Code Structure
• Graph-Based Characterization• Static Code Graphs• Dynamic Behavioral Graphs
…
Graphs
Static Graphs Dynamic Graphs
State-of-the-Art
Byte Entropy Summarized Code
1 0 1 1
![Page 9: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/9.jpg)
WWW.CYBER2020.COM
![Page 10: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/10.jpg)
The Solution: Deep Learning Applied to Cybersecurity
Deep Learning High-Performance Cloud Computing
Training Data Sets: Repository of
Billions of Malware
20 Years R&D
![Page 11: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/11.jpg)
WWW.CYBER2020.COM
![Page 12: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/12.jpg)
Cyber 20/20 is an all-software product consisting of:1) a deep learning neural network in the cloud
2) a network tap used to extract and examine all binary files entering an enterprise
3) a web based user interface
The Product
![Page 13: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/13.jpg)
WWW.CYBER2020.COM
![Page 14: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/14.jpg)
Great Start!
* Awarded $600K contract from Naval Research Laboratory.
* Awarded 1.25 Million compute hours on Department of Energy Titan supercomputer (fastest US supercomputer) to enhance our characterization and machine learning models.
* Applied for patent application No. 15256883, “Malware Analysis and Detection Using Graph-Based Characterization and Machine Learning”
![Page 15: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/15.jpg)
WWW.CYBER2020.COM
![Page 16: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/16.jpg)
Target Market Verticals
• Insurance and Financial Firms
• DoD, DHS, IC
• Critical infrastructure and healthcare
![Page 17: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/17.jpg)
WWW.CYBER2020.COM
![Page 18: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/18.jpg)
Competitive / CooperativeLandscape
![Page 19: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/19.jpg)
Barriers To Entry For CompetitionGraph-based
Characterization+
Machine Learning
• Patent pending for richest characterization of malware: graph-based characterization
• Know-how second to none: 20 years of code characterization and machine learning
• World-class experts in machine/deep learning, reverse engineering, high-performance computing, and malware analysis
![Page 20: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/20.jpg)
WWW.CYBER2020.COM
![Page 21: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/21.jpg)
Pricing and Projected RevenueRecurring Revenue Focus
Novel Pricing SchemeCost of Service : 1¢ per File
Projected RevenueProjected Profitability in Y2
Years
Organization Size
Monthly# of Files
Monthly Cost
Very Large 10 Million $100,000
Large 1 Million $10,000
Medium 250K $2500
Small 100K $1000 $ 0 Mil$ 10 Mil$ 20 Mil$ 30 Mil$ 40 Mil$ 50 Mil$ 60 Mil
2017 2018 2019 2020 2021
![Page 22: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/22.jpg)
WWW.CYBER2020.COM
![Page 23: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/23.jpg)
Capital Request
• 24-month runway• Early exit potential• Uses of funds
• Develop platform & product features
• Build sales and marketing engine$X.X Mil
![Page 24: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/24.jpg)
WWW.CYBER2020.COM
![Page 25: Cyber 2020 PitchDeck 091217 - eecis.udel.eduOver 100K malware variants created every hour *Bad actors have embraced automation *Good actors still construct malware detection rules](https://reader031.vdocuments.mx/reader031/viewer/2022022507/5acbad0b7f8b9ab10a8b8cb5/html5/thumbnails/25.jpg)
Deep Learning Meets Cybersecurity
John Cavazos, [email protected]
Jay Modi, [email protected]
WWW.CYBER2020.COM