Create a custom practice set
Pick category, difficulty, number of questions, and time limit. Start instantly with your own quiz.
Generate QuizGK Questions
Filter by category, type, and difficulty. Reading is open for everyone.
The Production Linked Incentive scheme for IT hardware aims to boost domestic manufacturing of laptops and servers.
- True
- False
Answer: True
PLI for IT hardware (2021) provides incentives on incremental sales of laptops, tablets, servers manufactured in India. Aims to reduce imports, create jobs, strengthen electronics ecosystem.
Data leakage occurs when test data influences training, leading to overly optimistic performance estimates.
- True
- False
Answer: True
Data leakage: using future data, target information, or test set in training/preprocessing. Causes overfitting and poor generalization. Critical to prevent via proper data splitting and pipeline design.
Multi-region deployment improves application availability by protecting against regional outages.
- True
- False
Answer: True
Multi-region architecture distributes workload across geographic regions, enabling failover during regional disasters. Critical for business continuity and global user experience.
GitOps uses Git repositories as single source of truth for infrastructure and ML deployments.
- True
- False
Answer: True
GitOps (ArgoCD, Flux) declaratively defines desired state in Git; operators sync cluster state. Benefits: audit trail, rollback via git revert, peer review via PRs. Critical for cloud-native MLOps.
Self-supervised learning enables models to learn from unlabeled data by predicting masked tokens.
- True
- False
Answer: True
Self-supervised learning (BERT, GPT pre-training) creates supervisory signals from data structure: mask prediction, next sentence prediction. Enables learning from vast unlabeled corpora.
The National Education Policy 2020 mandates integration of coding and AI literacy from middle school.
- True
- False
Answer: True
NEP 2020 emphasizes computational thinking, coding, AI exposure from Class 6 onwards, with experiential learning and teacher training. Implemented via DIKSHA, CBSE updates, ATLs.
Cross-validation provides more reliable model evaluation than single train-test split.
- True
- False
Answer: True
Cross-validation (k-fold) trains/evaluates on multiple data splits, reducing variance in performance estimates. Critical for small datasets and reliable model selection.
Infrastructure as Code scanning tools detect security misconfigurations before deployment.
- True
- False
Answer: True
IaC scanners (Checkov, tfsec) analyze Terraform/CloudFormation for misconfigurations: public storage, weak IAM, unencrypted resources. Critical for shift-left security in DevOps.
Model registries track versions, metadata, and lineage of ML models for governance.
- True
- False
Answer: True
Model registries (MLflow, SageMaker) manage model versions, parameters, metrics, and deployment history. Enables auditability, reproducibility, and rollback. Critical for enterprise MLOps.
Knowledge distillation transfers capabilities from large teacher models to smaller student models.
- True
- False
Answer: True
Knowledge distillation trains compact models to mimic large teacher outputs, enabling efficient deployment on edge devices with minimal accuracy loss. Critical for scalable AI systems.
The Digital Personal Data Protection Act 2023 applies extraterritorially to processing related to Indian individuals.
- True
- False
Answer: True
DPDP Act governs processing of digital personal data: (1) within India, and (2) outside India if offering goods/services or profiling individuals in India. Aligns with GDPR's extraterritorial scope.
Feature engineering creates new input variables from raw data to improve model performance.
- True
- False
Answer: True
Feature engineering: domain knowledge, transformations, interactions, encoding. Often more impactful than algorithm selection. Critical for successful machine learning projects.
Service mesh provides infrastructure-level observability, security, and traffic management for microservices.
- True
- False
Answer: True
Service mesh (Istio, Linkerd) deploys sidecar proxies for mTLS, retries, metrics, tracing. Decouples infrastructure concerns from application code. Critical for cloud-native architectures.
Canary deployments release new model versions to small user subsets before full rollout.
- True
- False
Answer: True
Canary deployment minimizes risk by gradually expanding model exposure, monitoring metrics, and rolling back if issues arise. Critical for safe ML system updates in production.
LoRA (Low-Rank Adaptation) fine-tunes large models by updating small rank decomposition matrices.
- True
- False
Answer: True
LoRA adds low-rank matrices to weight updates, reducing trainable parameters by 100-1000x vs full fine-tuning. Enables efficient adaptation of LLMs on consumer hardware. Critical for resource-constrained AI.
Data localization requirements mandate storing certain data within national borders for sovereignty.
- True
- False
Answer: True
Data localization (RBI payment data, DPDP Act provisions) ensures law enforcement access, reduces foreign surveillance risks. Balances sovereignty with global data flow needs. Critical for policy questions.
Explainable AI techniques help stakeholders understand model predictions for trust and compliance.
- True
- False
Answer: True
XAI methods (SHAP, LIME, attention visualization) provide interpretable explanations for AI decisions. Critical for regulated sectors (finance, healthcare) and user acceptance.
Serverless computing charges based on actual execution time rather than provisioned capacity.
- True
- False
Answer: True
Serverless (Lambda, Cloud Functions) uses pay-per-execution pricing, auto-scaling from zero. Trade-offs: cold starts, execution limits, vendor lock-in. Critical for cost-optimized cloud architectures.
Deception technology uses honeypots to detect and study attacker behavior.
- True
- False
Answer: True
Deception deploys fake assets (servers, credentials) to lure attackers, enabling early detection and threat intelligence. Critical for proactive defense against advanced adversaries.
Feature stores enable consistent feature computation across training and serving pipelines.
- True
- False
Answer: True
Feature stores (Feast, Tecton) manage feature engineering, versioning, and serving, preventing training-serving skew. Critical for reliable ML system performance in production.