Guardado en:
| Autores principales: | , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | https://arxiv.org/abs/2603.13453 |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866914392540446720 |
|---|---|
| author | Adeel, Ahsan Bilal, M. |
| author_facet | Adeel, Ahsan Bilal, M. |
| contents | Drawing on recent breakthroughs in cellular neurobiology and detailed biophysical modeling linking neocortical pyramidal neurons to distinct mental-state regimes, this work introduces a mathematically grounded formulation showing how models (e.g., Transformers) can implement computational principles underlying awake imaginative thought to pre-select relevant information before attention is applied via triadic modulation loops among queries ($Q$), keys ($K$), and values ($V$).~Scalability experiments on ImageNet-1K, benchmarked against a standard Vision Transformer (ViT), demonstrate significantly faster learning with reduced computational demand (fewer heads, layers, and tokens), consistent with our prior findings in reinforcement learning and language modeling. The approach operates at approximately $\mathcal{O}(N)$ complexity with respect to the number of input tokens $N$. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_13453 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Scalable Machines with Intrinsic Higher Mental-State Dynamics Adeel, Ahsan Bilal, M. Machine Learning Artificial Intelligence Drawing on recent breakthroughs in cellular neurobiology and detailed biophysical modeling linking neocortical pyramidal neurons to distinct mental-state regimes, this work introduces a mathematically grounded formulation showing how models (e.g., Transformers) can implement computational principles underlying awake imaginative thought to pre-select relevant information before attention is applied via triadic modulation loops among queries ($Q$), keys ($K$), and values ($V$).~Scalability experiments on ImageNet-1K, benchmarked against a standard Vision Transformer (ViT), demonstrate significantly faster learning with reduced computational demand (fewer heads, layers, and tokens), consistent with our prior findings in reinforcement learning and language modeling. The approach operates at approximately $\mathcal{O}(N)$ complexity with respect to the number of input tokens $N$. |
| title | Scalable Machines with Intrinsic Higher Mental-State Dynamics |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2603.13453 |