Evolution of Recurrent Neural Networks to Control Autonomous Life Agents
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Studies of artificial life (alife) attempt to simulate simple living beings. On the other hand, autonomous agents researchers are interested in building agents that are able to complete a particular task without supervision. In this research, these two areas of artificial intelligence are combined together into what we call "Autonomous Life Agent" (ALA). ALA is an artificial agent that is sent to some environment to live in without any supervision or any predefined behaviour rules. The primary goal of the agent is to learn how to survive in the artificial environment it lives in. In this research, we utilize a recurrent neural network to determine the agent's actions. A novel ALA Training System was developed that evolves recurrent neural networks using genetic algorithms. The resulting agents are capable of living in multiple similar worlds starting from random initial positions as well as in worlds that are unseen during the training. ACKNOWLEDGEMENTS Special thanks and appreciation to Dr. Jianna Zhang, Dr. Nick Cercone, and Dr. Brian Ross for their valuable comments The financial support for this project was provided by the Natural Sciences and Engineering Research Council of Canada (NSERC) under the USRA award, NSERC Grant 228142-2000, and by Communications and Information Technology Ontario (CITO). TABLE OF CONTENTS ABSTRACT ..............................................................................................................................2 ACKNOWLEDGEMENTS ......................................................................................................3 TABLE OF CONTENTS ..........................................................................................................4 CHAPTER 1:
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it