3.2 AI Network Design Use Cases (such as Machine Learning, Large Language Models, and Pattern Recognition) 400-007 Practice Questions (Page 4)

Questions 16–20

1. 16foundation · easy

   Which security control is most directly used to protect the integrity of a machine learning model during training?

   • AIncreasing the MTU on the training network
   • BNetwork segmentation to isolate the training environment
   • CUsing a single flat network to simplify monitoring
   • DDisabling encryption to reduce latency
   Reveal Solution Discussion Select an answer first
2. 17application · medium

   A government agency is deploying a pattern recognition system for facial recognition at border control points. The system uses edge devices at each checkpoint that send encrypted biometric data to a central processing center. The agency requires that the biometric data be protected from unauthorized access during transmission and that the edge devices be managed securely. Which network design approach best meets these requirements?

   • AUse IPsec VPN tunnels from each edge device to the central processing center and manage the devices via a secure management network
   • BSend the biometric data over the public internet without encryption and use a firewall at the central processing center
   • CUse a dedicated leased line from each edge device to the central processing center and manage the devices out-of-band
   • DUse MPLS VPNs to connect the edge devices to the central processing center and rely on the service provider for security
   Reveal Solution Discussion Select an answer first
4. 18application · medium

   A manufacturing company is deploying a pattern recognition system for predictive maintenance on its factory floor. The system uses sensors that send data to a central server for analysis. The analysis must be performed in near real-time to trigger maintenance alerts. The factory has a highly reliable but bandwidth-constrained network. Which design approach best meets the requirements?

   • ASend all sensor data to the central server and use a high-performance analytics platform to process it
   • BUse a store-and-forward approach where sensor data is batched and sent to the central server every hour
   • CDeploy edge processing nodes on the factory floor to analyze sensor data locally and send only alerts to the central server
   • DIncrease the bandwidth of the network to support sending all sensor data to the central server
   Reveal Solution Discussion Select an answer first
5. 19application · medium

   A retail company is deploying a real-time pattern recognition system for customer behavior analysis in its stores. The system uses cameras and edge inference nodes in each store to analyze video and send only anonymized insights to a central cloud. The company wants to minimize the cost of the WAN connectivity. Which design consideration is most important?

   • AImplement a content delivery network (CDN) to cache the video streams at the edge
   • BUse a dedicated high-bandwidth link from each store to the central cloud to ensure real-time delivery of insights
   • CDesign the WAN to support low-latency, reliable delivery of the small, time-sensitive insights to the central cloud
   • DEnsure the WAN links have sufficient bandwidth to support the continuous video stream from each store
   Reveal Solution Discussion Select an answer first
6. 20application · medium

   A smart-city project deploys a pattern recognition system for real-time video analytics at 500 traffic intersections. Each intersection runs an edge inference node that processes video locally and sends only anomaly alerts and metadata to a central data center. The central data center runs a model retraining pipeline. The network team must design the WAN connectivity between the edge nodes and the data center. Which design consideration is most important for this use case?

   • AGuaranteed bandwidth for each edge node to ensure the retraining pipeline receives all data in real time
   • BLow-latency, reliable connectivity for the small, time-sensitive alert traffic, with sufficient bandwidth for periodic model updates
   • CHigh-bandwidth, low-latency links between each edge node and the data center to support continuous video streaming
   • DA dedicated VPN tunnel from each intersection to the data center to secure the video stream
   Reveal Solution Discussion Select an answer first