Key details for this exam, checked against the published exam outline
Each question shows the correct answer and an explanation of why it is right
A retail company wants to build a system that can look at photos of products and automatically identify whether the item is a shirt, shoe, or bag, without a human manually labeling each pixel. The data scientist decides to use a neural network architecture that is particularly well-suited for extracting spatial features from images, such as edges and textures, through a series of filters.
Which type of deep learning architecture is best suited for this image classification task?
CNNs use convolutional filters/kernels that slide across an image to detect spatial patterns like edges, textures, and shapes, making them the standard architecture for image classification tasks. RNNs and LSTMs are designed for sequential data such as text or time series, not spatial image data. Reinforcement learning is used for sequential decision-making with rewards, not for static image classification.
A bank wants to build a model that predicts whether a loan applicant will default (yes/no) based on historical data that includes income, credit score, and repayment history, all of which are labeled with the actual outcome (defaulted or not).
Which category of machine learning best describes this use case?
Because the historical data includes labeled outcomes (default or no default) and the target variable is a discrete category, this is a supervised classification problem. Regression would apply if the target were a continuous numeric value (e.g., predicting a loan amount). Clustering is unsupervised and does not use labeled outcomes. Reinforcement learning involves an agent learning through rewards from interacting with an environment, which does not apply here.
A development team is building a chatbot application powered by a large language model (LLM). They want the model to understand long-range dependencies between words in a sentence and process the entire input sequence in parallel rather than sequentially, using a mechanism that allows the model to weigh the importance of different words relative to each other regardless of their distance in the text.
Which architectural component is primarily responsible for this capability in modern LLMs?
The self-attention mechanism, central to the Transformer architecture, allows the model to weigh relationships between all words in a sequence simultaneously (in parallel), capturing long-range dependencies without the sequential bottleneck found in RNNs/LSTMs. Convolutional pooling is used in CNNs for spatial feature reduction, not sequence understanding. Recurrent hidden states are used in RNNs/LSTMs but process data sequentially, which is slower and harder to parallelize. K-means clustering is an unsupervised learning algorithm, unrelated to Transformer architecture.
An enterprise wants to deploy an OCI Generative AI-powered application that can query an Oracle Autonomous Database using natural language questions instead of writing SQL manually, while also using vector search to retrieve semantically similar documents stored in the same database to ground the LLM's responses.
Which combination of Oracle 23ai capabilities enables this scenario?
Select AI, a feature of Oracle Autonomous Database (23ai), converts natural language prompts into SQL queries by leveraging generative AI. Oracle AI Vector Search adds native vector data types and similarity search directly in the database, enabling retrieval-augmented generation (RAG) style grounding of LLM responses. The other options are unrelated services: Data Science notebooks do not natively translate NL to SQL, Kubernetes is a container orchestration platform, Object Storage lifecycle policies manage data retention, and Vision API is for image analysis, not SQL parsing.
A healthcare provider is building an application that must extract structured data (such as patient name, date of birth, and diagnosis codes) from scanned PDF intake forms, and also analyze the sentiment of patient feedback surveys. The provider wants to use pre-built OCI AI services via REST APIs rather than training custom models from scratch.
Which pair of OCI AI services should the provider use for these two tasks respectively?
OCI Document Understanding is purpose-built to extract text, key-value pairs, tables, and structured data from documents like forms and invoices, including handling handwriting and specific document types. OCI Language provides pre-trained capabilities including sentiment analysis, key phrase extraction, and named entity recognition on text data such as survey responses. OCI Vision is for image classification/object detection, OCI Speech converts speech to text, and OCI Anomaly Detection identifies anomalies in time-series data, none of which match these specific tasks correctly in the other options.
46 questions covering all exam domains, starting from $20
Exam domains verified against: Official Oracle 1Z0-1122-26 exam guide, last checked September 2026.
AI basics, AI applications, types of data. Understand the distinction between AI, machine learning and deep learning and how they relate to business problems.
Supervised learning with regression and classification, unsupervised learning, reinforcement learning fundamentals. Model evaluation and the bias-variance tradeoff are core concepts.
Sample question from this domain above: Q2
Deep learning fundamentals, convolutional neural networks for image recognition, recurrent and LSTM networks for sequence processing. Backpropagation and activation functions underpin these architectures.
Sample question from this domain above: Q1
Generative AI overview, large language model fundamentals, Transformer architecture and self-attention mechanisms. Prompt engineering, instruction tuning and fine-tuning techniques for LLMs.
Sample question from this domain above: Q3
OCI AI, ML and AI infrastructure services overview. Responsible AI principles and ethical considerations in deploying AI systems on the Oracle Cloud.
OCI Generative AI services, Autonomous Database AI features, Oracle Vector Search capabilities. Integration of generative AI into modern database systems.
Sample question from this domain above: Q4
OCI Language, Vision, Document Understanding and Speech services. Practical APIs and hands-on understanding of how these services solve real-world business problems.
Sample question from this domain above: Q5
Common questions about the exam itself
