MIS 375 Exam 1 UPDATED ACTUAL Questions And Correct Answers
Terms in this set (65)
Neural network Input layer, hidden layers, output layer
Vector database Stores unstructured data (text, images, audio) as embeddings for each
token/chunk
Training methods RAG and fine-tuning
Token Small unit of text/data that an AI model uses to interpret input and generate
output
Parameters Elements (numbers) within a vector database the model adjusts through training to
learn and make predictions
Frontier models General knowledge (i.e., Claude, Preplexity, Deepseek, Copilot)
Multimodal models Text, audio, and images abilities in the same model
Attention Mechanism that assigns importance weights to different parts of the input so a
model can selectively focus on the most relevant info when making predictions
(such as relationships across tokens)
Transformer Neural network architecture that processes all parts of a sequence in parallel
using attention, enabling performance across multi-modal tasks
Multi-layer perceptron Elements within a neural network that the model adjusts through training to learn
and make predictions
Cosine similarity Used for calculating similarity scores (can change during training)
Embedding Vector representation of token invector space
Retrieval-Augmented Generation (RAG) Allows uploading of trusted data or proprietary info to your own local vector
space
Good strategy is a result of a process
Is financial forecasting alone a strategy? No
Strategy Process of gathering ideas to create progress
Porter's 4 Generic Strategies cost leadership, differentiation, cost focus, focused differentiation
Terms in this set (65)
Neural network Input layer, hidden layers, output layer
Vector database Stores unstructured data (text, images, audio) as embeddings for each
token/chunk
Training methods RAG and fine-tuning
Token Small unit of text/data that an AI model uses to interpret input and generate
output
Parameters Elements (numbers) within a vector database the model adjusts through training to
learn and make predictions
Frontier models General knowledge (i.e., Claude, Preplexity, Deepseek, Copilot)
Multimodal models Text, audio, and images abilities in the same model
Attention Mechanism that assigns importance weights to different parts of the input so a
model can selectively focus on the most relevant info when making predictions
(such as relationships across tokens)
Transformer Neural network architecture that processes all parts of a sequence in parallel
using attention, enabling performance across multi-modal tasks
Multi-layer perceptron Elements within a neural network that the model adjusts through training to learn
and make predictions
Cosine similarity Used for calculating similarity scores (can change during training)
Embedding Vector representation of token invector space
Retrieval-Augmented Generation (RAG) Allows uploading of trusted data or proprietary info to your own local vector
space
Good strategy is a result of a process
Is financial forecasting alone a strategy? No
Strategy Process of gathering ideas to create progress
Porter's 4 Generic Strategies cost leadership, differentiation, cost focus, focused differentiation
