CSCI-B 651 Natural Language Processing
3 credits
- Prerequisite(s): CSCI-B 551 or CSCI-B 555 or CSCI-B 565 or INFO-H 515
- Delivery: On-Campus
- Semesters offered: Spring (Check the schedule to confirm.)
- Equivalent(s): CSCI 59000 Natural Language Processing
Description
Theory and methods for natural language processing. Algorithms for sentence parsing and generation. Context-free and unification grammars. Question-and-answer systems. Analysis of narratives. Finite-state approaches to computational phonology and morphology. Machine translation. Machine learning of natural language. Speech recognition. Neural network and statistical alternatives to symbolic approaches.
Topics
Natural language processing (NLP) foundations
- Significance in AI and real-world applications
- Text processing and foundational grammars
Language models
- Basics
- Word embeddings
- Pre-trained models
Text classification and sequence labeling
- Traditional and neural methods
Common neural network architectures for NLP
- Encoder–decoder
- Attention mechanisms
- LSTM
- Transformer
Semantics
- Analysis and implementation
Parsing and discourse analysis
- Coherence techniques
- Syntactic understanding
Applications
- Machine translation
- Text summarization
Ethical considerations
- Biases
- Privacy
- Societal implications
Independent research project
- Research design
- Literature review
- Implementation
- Analysis
Learning Outcomes
- Implement NLP techniques, such as text classification and sequence labeling, in solving real-world problems, demonstrating proficiency in addressing language-related tasks across different domains. CS 1
- Deconstruct and analyze complex NLP models, such as transformer-based neural networks, to discern the underlying mechanisms contributing to their effectiveness in various language tasks. CS 1
- Evaluate the suitability of different language models and pre-trained embeddings for specific NLP applications, considering model performance, efficiency, and adaptability across domains. CS 4
- Develop and refine NLP pipelines, incorporating pre-trained language models and transfer learning, to achieve high-quality results on specialized tasks like sentiment analysis or medical text comprehension. CS 2
- Implement syntactic and semantic analysis of text, including parsing and discourse coherence, enhancing the understanding of the structure and meaning of language. CS 1
- Implement techniques for downstream NLP applications such as automatic text summarization, condensing large volumes of information into coherent and concise summaries, exhibiting content abstraction. CS 4
- Assess the ethical implications of employing NLP techniques, including potential biases in language models and the impact of NLP technologies on privacy and other societal issues. CS 6
- Plan and execute independent research projects that extend the boundaries of NLP knowledge, applying innovative methodologies to address emerging challenges in the field and producing meaningful contributions to the research community. CS 7
Policies and Procedures
Please be aware of the following linked policies and procedures. Note that in individual courses instructors will have stipulations specific to their course.