5/4/26 WORKING ARCHITECTURE NEED TO SETUP BASIC ENVIRONMENT
Meeting for FPGA with Ryan, Taichi discussed a proposed machine learning framework using the judicial system and got approved to begin setting up environment. Trying to restrict languages to Verilog, Python, VHDL
5/9/26 FRAMEWORK AND LITERATURE REVIEW
Taichi and I met to plan out framework of ai model planned framework but need further insight for putting the model to run completely independently on a FPGA not just computing. Also, started a literature review to explore previous machine learning research.
https://docs.google.com/spreadsheets/d/1VF4O2ZNt870OuzxEBXH8aYwfNQ1nEMtRXLQjjfgCbKE/edit?usp=sharing
Main problems to overcome:
FPGA logic is hardware, not software — you must build scheduling, memory management, communication, mutation control, and supervision yourself in Verilog/VHDL.
Verilog/VHDL are static hardware-description languages, not dynamic self-modifying languages; generating valid new hardware safely at runtime is difficult.
Reconfigured regions reset and lose internal state unless memory/state is externally preserved.
5/10/26 SETUP CONSTITUTION IN VHDEL
Installed FPGA languages along with Constitution environment testing on hardware level with basic constitution for whether system outputs violate fixed safety rules-such as arithmetic overflow, illegal hardware states, unauthorized access requests, out-of-scope operations, invalid authorization, or outputs outside approved numeric bounds—before allowing them to propagate.
https://github.com/Toshiyuki037/FPGA-NN-MODELING/tree/main/constitution_layer
Installers Below:
https://github.com/ghdl/ghdl/releases?utm_source=chatgpt.com
https://bleyer.org/icarus/
https://liquidinstruments.com/moku-compile/
5/14/26 ZOOM RESEARCH MEETING
Met with team over zoom discussed model future problems, claims, and scopes of model. We are claiming that all processing can be done on a fpga so we need to look into a converter model. Also we were recommended to use AI to commit to a deep search on FPGA & Neural Network to see all existing research. Discussed if able to isolate neural network completely to FPGA we would need to build memory, CPU, etc using the transistors onto FPGA. In doing so, essential creating an an “artificial organism”.