Summary of "WEBINAR NASIONAL: Integrasi ML dan Big Data: Membangun Ekosistem Kecerdasan Digital yang Adaptif"

Summary of the Webinar

Title: WEBINAR NASIONAL: Integrasi ML dan Big Data: Membangun Ekosistem Kecerdasan Digital yang Adaptif

This national webinar, held on October 14, 2025, was organized by the Undergraduate Study Program of Information Systems at University of Science and Technology Semarang (STEKOM). It featured 10 expert speakers from various universities and institutions across Indonesia. The theme focused on the integration of Machine Learning (ML) and Big Data to build an adaptive and innovative digital intelligence ecosystem.

Main Ideas, Concepts, and Lessons Conveyed

1. Opening and Context Setting

Emphasized the rapid development of digital technologies, especially ML and Big Data, and their transformative impact on industries, education, government, and social sectors.
Highlighted the core challenge of effectively integrating ML and Big Data to create ecosystems adaptive to fast changes and innovation.

2. Fundamental Research as the Foundation (Mr. Dwi Atmojo WP)

Digital intelligence ecosystems consist of multiple layers:
- Infrastructure (cloud, 5G)
- Data & intelligence (big data, AI, ML)
- Platforms & services
- Human interaction (UI/UX, AR/VR)
- Governance and security
Fundamental/basic research is crucial to develop core algorithms and innovations fueling the ecosystem.
Local wisdom and natural phenomena can inspire new ML algorithms.
Research strengthens national capabilities and reduces dependence on foreign technology.

3. Machine Learning in Education (Mrs. Eka Yulasari)

ML personalizes online learning platforms by adapting to students’ unique learning styles (visual, auditory, kinesthetic).
Benefits include personalized learning paths, early detection of learning difficulties, automated feedback, and pedagogical support for teachers.
ML models used: classification, clustering, and recommendation systems.
Future adaptive learning platforms will dynamically adjust curriculum, content, and evaluations based on real-time student performance.

4. Applications of ML and Deep Learning in Medicine (Dr. Faisal Muttaqin)

AI assists doctors in clinical decision-making by analyzing large-scale medical data (X-ray, MRI, USG).
ML methods include supervised, unsupervised, and reinforcement learning; Deep Learning uses multi-layer neural networks (CNN, UNET, ResNet).
Deep learning excels in complex tasks like image recognition but requires large datasets and high computing power.
Medical data requires ethical clearance, anonymization, accurate labeling by experts, and data standardization.
Challenges: data availability, hardware needs, interpretability, and trust.
AI is a strategic partner, not a replacement for doctors.

5. Reinforcement Learning and Its Dynamics (Mr. Handoko Supeno)

Reinforcement learning (RL) is a paradigm where an agent learns by interacting with an environment and receiving rewards or penalties.
RL is akin to training pets, learning optimal actions through trial and error.
Deep Reinforcement Learning combines RL with deep neural networks for complex tasks like autonomous driving.
RL allows continuous learning and decision-making in uncertain or dynamic environments.

6. Maintaining the Human Role Amid AI Automation (Mr. Eko Siswanto)

AI-powered automation improves efficiency, accuracy, and accelerates processes but cannot replace uniquely human traits.
Humans excel in emotional intelligence, creativity, ethical decision-making, and social skills.
AI should be viewed as a tool to augment human productivity, not replace humans.
Continuous learning, skill development, and ethical governance are essential.
Education and policy must adapt curricula to prepare humans for collaboration with AI.

7. From Prediction to Prescription: Adaptive Ecosystems (Mrs. Herlina Jayadianti)

Many ML applications stop at prediction (e.g., sales forecasting, sentiment analysis).
The next step is prescriptive analytics—systems that recommend or automate decisions based on predictions.
Examples include:
- Smart grids adjusting power distribution
- Smart farming advising actions based on weather forecasts
- Adaptive traffic routing
Prescriptive systems enable faster, more accurate decision-making and operational efficiency.

8. Integration of Big Data and ML for Adaptive Ecosystems (Mrs. Resti Andriani)

Digital transformation involves adopting technology to improve productivity, customer interaction, and innovation.
Big Data characterized by 4Vs (Volume, Velocity, Variety, Veracity) plus Value.
ML enables efficient analysis of Big Data, supporting automated, data-driven decisions.
Integration creates ecosystems that adapt in real-time to market or user changes.
Challenges include data volume, quality, privacy, security, and resource limitations.

9. Fusion of Intelligence: ML + Big Data in Modern Digital Ecosystems (Mr. Agung Widodo)

ML is a subset of AI; Deep Learning is a subset of ML.
ML algorithms include regression, classification, clustering; Deep Learning uses neural networks.
Big Data’s 5Vs (Volume, Velocity, Variety, Veracity, Value) require platforms like Hadoop and Spark.
The synergy of ML and Big Data enables smart city applications, digital health, and other adaptive intelligent systems.
Challenges: data privacy, algorithmic bias, computing scalability, and limited human resources.
Solutions: governance frameworks, explainable AI, cross-sector collaboration, and emerging trends like federated learning and synthetic data.

10. National Research Agenda & Roadmap (Prof. Dr. Denrias Utomo)

Data explosion driven by increased users and devices necessitates real-time decision-making supported by AI.
ML and Big Data are key to automation, prediction, recommendation, and resource optimization.
National roadmap (2025-2030) focuses on downstreaming AI, IoT, and robotics in smart farming, digital health, logistics, manufacturing, and adaptive education.
Emphasis on infrastructure (5G, edge computing), standardization, open source, and AI literacy.
Collaboration between academia, industry, and government is crucial for impactful innovation.

11. Security in Machine Learning Lifecycle (Mr. Ali Hafiz)

ML and AI security is critical due to widespread adoption in sensitive sectors (business, education, government, health).
Risks include data poisoning, model inversion, backdoor attacks, and drift attacks.
Many organizations lack adequate AI security mechanisms.
Secure by design principle: embed security from the start of the ML lifecycle (data collection, preparation, training, evaluation, deployment, monitoring).
Use frameworks and standards like NIST, ISO 27001, and OWASP ML Security Top 10.
Strategies include data pipeline security, model hardening, security testing, audits, and governance.
Collaboration between AI, cybersecurity, and governance is essential.

Methodologies or Instructional Lists Presented

Digital Intelligence Ecosystem Layers (Mr. Dwi Atmojo)

Infrastructure (cloud, 5G)
Data & Intelligence (big data, AI, ML)
Platforms & Services (applications, public services)
Human Interaction (UI/UX, AR/VR)
Governance & Security (regulations, privacy)
Stakeholders (government, business, academia, society)

Adaptive Online Learning System Architecture (Mrs. Eka Yulasari)

Data collection (student activities)
Data preprocessing (cleaning)
Machine learning model (predict performance, difficulties)
Recommendation engine (tailored material)
Feedback loop (automatic feedback, NLP/LLM)

Reinforcement Learning Components (Mr. Handoko Supeno)

Agent (AI model)
Environment (interaction context)
State/Observation (environmental condition)
Action (agent’s response)
Reward (feedback signal)

Machine Learning Life Cycle Security (Mr. Ali Hafiz)

Data collection (risks: poisoning, theft)
Data preparation (risks: backdoor injection)
Model training & evaluation (risks: adversarial attacks)
Deployment (risks: model extraction)
Monitoring & maintenance (risks: drift attacks)
Mitigation: secure design, privilege management, layered defense, continuous threat modeling

Big Data & ML Integration Process (Mrs. Resti Andriani)

Data collection from diverse sources
Data cleaning and validation
ML model training and pattern finding
Prediction or automated decision-making
Real-time adaptation and response

Speakers/Sources Featured

Ika Puspasari – Moderator, Student at University Tekom Semarang
Mr. Dwi Atmojo WP, MKom – Lecturer, Perbanas Institute Jakarta
Mrs. Eka Yulasari, SKOM, MKom – Lecturer, Sarjanawiata University Yogyakarta
Dr. Faisal Muttaqin, SKOM, MT – Lecturer, UPN Veteran East Java Surabaya
Mr. Handoko Supeno, ST, MT – Lecturer, Pasudan University Bandung
Mr. Eko Siswanto, MKom – Lecturer, Tekom University Semarang
Mrs. Dr. Herlina Jayadi, ST, MT – Lecturer, UPN Veteran Yogyakarta
Mrs. Resti Andriani, S., MKom – Lecturer, Bima Sakapenta University Tegal
Mr. Agung Mulyo Widodo, ST, MSC, PhD – Lecturer, ESA University Yogyakarta & Researcher, Taiwan
Prof. Dr. Denrias Utomo, S.Si., MT – Lecturer, Jember State Polytechnic
Mr. Ali Hafiz, SKOM, MTI, CSA – Lecturer, Institute of Business Technology and Language Dian Cipta Cendekia, Lampung

Overall Conclusion

The webinar provided a comprehensive overview of how the integration of Machine Learning and Big Data is essential to building adaptive, innovative digital intelligence ecosystems across multiple sectors. Key components discussed included fundamental research, education, healthcare, reinforcement learning, human-AI collaboration, prescriptive analytics, and security.

The importance of national research agendas, infrastructure readiness, and governance frameworks was emphasized to ensure sustainable and impactful development. Despite increasing AI automation, the human role remains central in ethical, creative, and strategic decision-making.