Deep Neural Networks projects

Wed, 15 Jan 2025 00:00:00 +0000

This is a collection of projects completed for the Deep Neural Networks course, including:

Multilayer Perceptrons:
- Implementation of ProxyProp (Proximal Backpropagation) algorithm described in this paper.
Computer Vision:
- Implementation of several transformations, AdvProp, and SparseTopK methods.
Natural Language Processing:
- Implementation of SwiGLU and Differential Attention for a decoder-only Transformer architecture.
Reinforcement Learning:
- Implementation of Proximal Policy Optimization (PPO) and Random Network Distillation (RND) for MiniHack environments.

Logic in Computer Science

Fri, 24 May 2024 00:00:00 +0000

This is a collection of projects completed for the Logic in Computer Science course, including:

Propositional logic in Haskell (Part 1):
- Satisfiable formulas, tautologies, disjunctive normal form (DNF), DNF-based SAT solver.
Propositional logic in Haskell (Part 2):
- Conjunctive normal form (CNF), equisatisfiable CNF (ENCF), Davis-Putnam SAT solver.
SAT solving with Z3 in Python (Part 1):
- Generating all satisfying assignments, pebbling formulas, n-queens problem.
SAT solving with Z3 in Python (Part 2):
- Shortest common superstring (SCS) problem, phase transitions for the k-SAT problem.
SMT solving with Z3 in Python:
- Łoś principle, function 91, Sudoku solver.
Intuitionistic propositional logic in Agda:
- Conjunction and disjunction proofs, de Morgan’s laws, weak Pierce’s law.
Intuitionistic first order logic in Agda:
- Universal and existential quantifier proofs, relations.