Sequential Planning for Steering Immune System Adaptation
Authors: Christian Kroer, Tuomas Sandholm
IJCAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We run experiments with two alternate goals: developing regulatory T cells or developing effector T cells. |
| Researcher Affiliation | Academia | Computer Science Department Carnegie Mellon University Pittsburgh, PA, USA {ckroer,sandholm}@cs.cmu.edu |
| Pseudocode | Yes | Algorithm 1 UCT that leverages a (biology) simulator |
| Open Source Code | No | The paper mentions 'All the plans are available online' with a URL, but this refers to the generated plans, not the source code for the methodology itself. No explicit statement of source code release for their implementation is provided. |
| Open Datasets | No | The paper uses a T-cell simulator model (Bio Net Gen) which was calibrated by others with wet lab experiments. It does not describe training a model on a dataset or provide access information for such a dataset relevant to its own experiments. |
| Dataset Splits | No | The paper discusses the T-cell model's calibration ('calibrated with wet lab experiments') but does not specify validation splits for an experimental dataset used in their work. |
| Hardware Specification | No | The paper does not provide any specific details about the hardware used to run the experiments (e.g., CPU, GPU models, memory). |
| Software Dependencies | No | The paper mentions using 'Bio Net Gen' and the 'UCT algorithm' but does not specify version numbers for these or any other software dependencies directly within the text as required for reproducibility. |
| Experiment Setup | Yes | For each experiment, we ran UCT for 3000 iterations. At every 100th iteration, we took the current greedy policy and sampled an expected value by performing 100 rollouts of Bio Net Gen simulations with that strategy. We repeated each experiment 7 times, and report the average expected values over the 7 experiments. We experimented with four different simulation times t: 36, 72, 360, and 1080... We considered treatment plans of depth 1, 2, and 3. |