Attachment_X_Proposal_Overview_and_Proposed_Metrics.pptx

DARPA-PA-26-05 MATHBAC

Guidance – Delete after reading Instructions are provided on the slides in yellow with red text. Please delete them before submitting Proposers only need to fill out the information for their proposed Technical Area References and images are highly encouraged Do not try to pack the slides with too much information. Less is better.

Do not go below 11pt Tahoma (excluding references)

TA1: The Mathematics of Agentic Communication Protocols

DARPA-PA-26-05 MATHBAC

Instructions (delete before submitting) Include image of current agentic platform

How did you set your targets?

Why are they challenging?

Which “small science models” do/will you use?

# of agents:

types of agents:

model(s) used:

# of tasks / domains you work in:

# of months it took to build (current) agentic platform:

Instructions (delete before submitting) See MATHBAC ES “Technical Area 1” for additional context Proposers are expected to identify the mathematics, systems theory, and information theory methods that will be leveraged and advanced to understand/design and improve the communication protocols of interacting agents. These may include input-output attempts to create for each agent approximate “internal models” of the agents with which it interacts (a sort of “theory of mind” of agents). Proposers must state:

(i) the methods that will be used to

(a) evaluate and schedule interactions between heterogeneous agents

(b) understand and enable more efficient (e.g., lower communication prompt length/faster completion) agent interactions (b1) without access to any latent spaces (b2) with access to transformer latent spaces (b3) with mixed access agents (with/without accessible latent spaces)

(c) characterize and quantify the progress of communication dynamics throughout the campaign

(d) optimize communication protocols.

Agentic Platform Proposed Mathematics/Systems/Information/Data Science Methods Domain and Tasks Metrics

TA2: Geometry and Evolution of Shared Knowledge – Generalizable Principles from Data

DARPA-PA-26-05 MATHBAC

How did you set your targets?

Why are they challenging?

& Corpora

Instructions (delete before submitting) See MATHBAC ES “Technical Area 2” for additional context Proposers are expected to identify the mathematics, information theory, and data science methods that they intend to leverage and advance towards discovering new generalizable principles by monitoring and analyzing multiple data streams. Proposals must describe:

Methods the proposer will use to

(a) enable combination/comparisons of data streams across modalities and time scales.

(b) identify common and uncommon/particular features of the data streams, e.g., experimental data streams or streams arising from agentic interactions.

(c) extract generalizable principles, laws, or correlations; explicitly describe cost function/rewards used to drive principle/law/correlation extraction.

(d) enable interpretability of the extracted principles, laws, or correlations.

(e) generate and validate a minimum curriculum (content) for succeeding in a given task/science domain, if possible.

What is your training corpora?

How are you ensuring these corpora do not explicitly contain nor purposefully seed your targeted principle(s), law(s), or correlation(s)?

Data Sources/Streams Used Proposed Mathematics Domain, Tasks, and proposed principles, laws, or correlations Metrics

DARPA-PA-26-05 MATHBAC

Proposed Metric # 1 Technical Area:

Description:

Method for calculation:

How does the metric measure progress against the program goals:

How does the metric supplement current program metrics:

Why should this metric be adopted? (e.g. what insights will it provide)

Literature sources providing metric provenance and application (if available) Each proposed metric is limited to one slide; duplicate the slide for each additional metric [These are instructions, delete after reading] image1.png image2.png