AI Glossary for Treasury
56 AI concepts explained in plain language, and a map of which AI technologies sit behind common treasury use cases. Keep it on your desk. Looking for bank connectivity, payment formats or controls? See the full Treasury Technology Glossary.
Core AI Concepts
- Artificial IntelligenceAI
- The field of computer science focused on creating systems capable of performing tasks that typically require human intelligence.
- Machine LearningML
- A subset of AI that enables computers to learn and improve from experience without being explicitly programmed for each task.
- Deep Learning
- A subset of ML that uses artificial neural networks with multiple layers to process complex data.
- Neural Networks
- Computational models inspired by the human brain, consisting of interconnected nodes that process information.
- Generative AI
- AI technology that can create new content (text, images, audio) based on training data.
- Large Language ModelsLLMs
- AI models trained on massive volumes of text to understand and generate natural language. Examples: GPT, Claude.
Types of Machine Learning
- Supervised Learning
- Learning with labeled data, where the model learns from examples with known answers.
- Unsupervised Learning
- Learning with unlabeled data, where the model discovers patterns and structures on its own.
- Reinforcement Learning
- Learning through trial and error, where the model receives rewards or penalties for its actions.
- Few-shot Learning
- Learning with very few training examples.
- Zero-shot Learning
- The ability to recognize categories never seen during training.
- Transfer Learning
- Using a pre-trained model for a new, related task.
Natural Language Processing
- NLPNatural Language Processing
- The field of AI dealing with interaction between computers and human language.
- NLUNatural Language Understanding
- The part of NLP focused on understanding the meaning of text.
- NLGNatural Language Generation
- The part of NLP focused on generating text in natural language.
- Tokenization
- The process of splitting text into smaller units (words, subwords) for model processing.
- Embeddings
- Numerical representations of words or phrases in multidimensional vector spaces.
- Sentiment Analysis
- Analyzing the sentiment or emotion expressed in text.
- Named Entity RecognitionNER
- Identifying entities in text such as people, places, organizations.
Key Architectures
- Transformers
- Neural network architecture that revolutionized NLP, based on the attention mechanism. Foundation of GPT, BERT, Claude.
- Attention Mechanism
- A technique allowing models to focus on relevant parts of input data.
- RNNsRecurrent Neural Networks
- Neural networks specialized for processing sequences of data.
- LSTMLong Short-Term Memory
- A special type of RNN capable of retaining information over long periods.
- Mixture of ExpertsMoE
- Architecture with specialized modules activated conditionally, enabling larger models with less compute.
- GANsGenerative Adversarial Networks
- Two neural networks competing to generate increasingly realistic content.
- Diffusion Models
- Generative models that create images by gradually removing noise from random data.
Technical Concepts
- Training Data
- The dataset used to train an AI model.
- Overfitting
- When a model learns the training data too well and fails to generalize to new data.
- Underfitting
- When a model is too simple to capture patterns in the data.
- Hyperparameters
- Configuration parameters of a model that are not learned automatically.
- Fine-tuning
- Adapting a pre-trained model to a specific task using additional training data.
- Prompt Engineering
- The practice of crafting effective inputs to get optimal responses from LLMs.
- RAGRetrieval-Augmented Generation
- Combining LLMs with external knowledge retrieval to improve accuracy and reduce hallucinations.
- Hallucination
- When an AI model generates plausible-sounding but factually incorrect information.
Applied AI & Automation
- RPARobotic Process Automation
- Automating business processes using software bots. Not AI, but often marketed as AI.
- OCROptical Character Recognition
- Recognizing and extracting text from images and scanned documents.
- Anomaly Detection
- Identifying unusual behaviors or data points that deviate from expected patterns.
- Time Series Forecasting
- Predicting future values based on historical time-ordered data.
- Chatbots
- Programs that simulate human conversation.
- Ensemble Methods
- Combining multiple models for improved performance. Includes Random Forest, Gradient Boosting.
Evaluation Metrics
- Accuracy
- The percentage of correct predictions out of total predictions.
- Precision
- The proportion of positive predictions that are actually correct.
- Recall
- The proportion of actual positive cases that are correctly identified.
- F1 Score
- The harmonic mean of precision and recall.
- Perplexity
- A measure of a language model’s uncertainty. Lower is better.
AI Infrastructure
- MLOps
- Practices for operationalizing machine learning models in production.
- Model Drift
- Degradation of model performance over time as real-world data changes.
- Data Pipeline
- An automated flow for processing and transforming data.
- A/B Testing
- Comparative testing of models under real conditions.
- Vector Database
- A database optimized for storing and querying embedding vectors, essential for RAG systems.
- Model Monitoring
- Continuous surveillance of model performance in production.
AI Safety & Ethics
- Explainable AIXAI
- Development of AI models whose decisions can be understood by humans.
- AI Ethics
- The study of moral and social implications of AI technologies.
- Federated Learning
- Training models on distributed data without centralizing it, preserving privacy.
- Adversarial Examples
- Data modified to fool AI models into making incorrect predictions.
- Privacy-Preserving AI
- Techniques for protecting data confidentiality while using AI.
No terms match that filter.
AI in Treasury: Technologies & Applications
Which AI techniques sit behind common treasury use cases, and examples of vendors and platforms operating in each area. Vendor names are examples, not recommendations; products and ownership change, so check each vendor directly.
Predictive Analytics & Forecasting
| Use case | Technologies | Example vendors |
|---|---|---|
| Cash Flow Forecasting | Time Series, LSTM, ARIMA, Prophet | Kyriba, Coupa Treasury (Bellin), SAP, Oracle |
| Liquidity Management | ML, Regression Models, Monte Carlo | ION Treasury (Reval), FIS Integrity |
| FX Rate Prediction | Deep Learning, RNNs, Transformers, Ensembles | Bloomberg Terminal, LSEG Workspace (Refinitiv Eikon) |
| Interest Rate Forecasting | Time Series, Bayesian Networks, Neural Networks | Moody’s Analytics, S&P Capital IQ |
Risk Management
| Use case | Technologies | Example vendors |
|---|---|---|
| Credit Risk Assessment | Supervised Learning, Random Forest, SVM | Moody’s RiskCalc, S&P CreditModel, FICO |
| Market Risk Analysis | Monte Carlo, VaR Models, ML | Murex, Calypso, FIS |
| Counterparty Risk | Classification, Ensemble Methods, Neural Networks | IBM OpenPages, Algorithmics |
| Operational Risk Detection | Anomaly Detection, Unsupervised Learning | SAS Risk Management |
Fraud Detection & Security
| Use case | Technologies | Example vendors |
|---|---|---|
| Transaction Fraud Detection | Real-time ML, Anomaly Detection, Graph Neural Networks | FICO Falcon, SAS Fraud Management |
| Payment Fraud Prevention | Deep Learning, Pattern Recognition | Featurespace ARIC, DataVisor |
| AML Compliance | NLP, Graph Analytics, Supervised Learning | NICE Actimize, Quantexa |
Process Automation
| Use case | Technologies | Example vendors |
|---|---|---|
| Invoice Processing & AP | OCR, NLP, Computer Vision, RPA | MindBridge, AppZen, Stampli |
| Bank Reconciliation | String Matching, Fuzzy Logic, ML | ReconArt, SmartStream, BlackLine |
| Regulatory Reporting | RPA, NLP, Data Mining | Wolters Kluwer, AxiomSL |
Built by a treasurer, for treasurers. · treasuryease.com