Add academic paper and citation: CITATION.md
Browse files- CITATION.md +27 -0
CITATION.md
ADDED
@@ -0,0 +1,27 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# Citation
|
2 |
+
|
3 |
+
If you use this work in your research, please cite:
|
4 |
+
|
5 |
+
```bibtex
|
6 |
+
@article{murray2025biotranscendent,
|
7 |
+
title={Mathematical Framework for Bio-Transcendent Intelligence: with Phase-4 Application Results},
|
8 |
+
author={Murray, Jeff N},
|
9 |
+
year={2025},
|
10 |
+
month={August},
|
11 |
+
note={Personal research project. Available at: https://huggingface.co/jmurray10/phase4-quantum-compression}
|
12 |
+
}
|
13 |
+
```
|
14 |
+
|
15 |
+
## Paper Abstract
|
16 |
+
|
17 |
+
We present an application-validated framework for intelligence beyond biological constraints, combining quantum search, adaptive computation, reliability-first memory, and efficiency-driven compression. New Phase-4 experiments demonstrate:
|
18 |
+
|
19 |
+
1. Grover's search executed end-to-end on real quantum hardware (IBM backend), achieving marked-state success of 59.9% over 2048 shots (simulator: 95.3%)
|
20 |
+
2. Near-theoretical 4� model compression on MLPs (3.92� measured from serialized weights) with minimal output drift (MAE 0.0018)
|
21 |
+
3. An energy proxy reduction of 62.1% for INT8 vs FP32 with a small latency trade-off on CPU small batches
|
22 |
+
4. Genuine optimization trade-offs where SGD is 3.9� faster while Evolution attains 128� lower final loss
|
23 |
+
|
24 |
+
## Access the Full Paper
|
25 |
+
|
26 |
+
- [Download PDF](./papers/BioTranscendent_Intelligence_Phase4.pdf)
|
27 |
+
- [View on GitHub](https://github.com/jmurray10/phse_4)
|