Can a verifier tell from old-task losses that a continual-learning job skipped or shortchanged tasks?

research/forgetting-audit

Auditing skipped or partial continual-learning training from old-task losses: exact loss law (ρ^{t−1}[(1−2m₁+m₂)(r/d)λᵏ+(r/d)ρ(ρᵏ−λᵏ)]), a skipped task is ≥ d/(d−r)× worse but only clearly so inside a lag window, detecting a skipper takes ~0.7/s² jobs at any lag, a 10% step shortfall needs ~9600.

7 tests

What it found

Extends the forgetting law to a trainer that skips each task with probability s or takes a step fraction a: the final-checkpoint loss on the task at position t is exactly ρ^t−1[(1−2m₁+m₂)(r/d)λ^k + (r/d)ρ(ρ^k−λ^k)], matching exact quadrature and simulation. A skipped task is at least d/(d−r) times worse than a trained one at every lag but clearly so only in a window (ratio ≥2 for lags ≤6.7 at d=16, r=2). Catching a skipping trainer is lag-free: about 17–20 jobs for 20% skipping at any audit lag, scaling as 0.7/s² and falling with job length, while a 10% step shortfall needs about 9,600 jobs. Stylised: linear regression, Haar-random tasks, known initial error.

A figure from its own output

Figure produced by research/forgetting-audit: skip_ratio — forgetting-audit/experiments/skip_ratio.svg, written by the project's own experiment code.

Paper, code and results

More in Detection & tolerances

Is an interference-free or mean-noise-rise radar twin right when robots' radars interfere with each other?Interference-free or mean-noise-rise radar twin vs intermittent mutual radar interference between robots: exact closed-form per-frame Pd, real Pd plateaus at 0.656 (no-hit probability), mean-rise twin 9% optimistic at Pd ≥ 0.9 yet badly pessimistic for hit-tolerant missions, geometry knowledge helps 9-of-10 missions (Brier 0.089 vs 0.101). Stylised.
Does demand-driven simulation mislead leak detection on a stressed water network, and where does pressure-driven demand change the answer?Net3 leak signals under demand-driven and pressure-driven demand (WNTR): identical at 1 to 1.25 times base demand and nearly so at 1.5 times; beyond that both models fail because the pumps cannot meet demand (simulated; one network).
How small a water-main leak can pressure sensors find, and does a changepoint detector beat a threshold or CUSUM?Water-leak detection from pressure residuals on EPANET's Net3 (WNTR): at a 5% weekly false-alarm budget, 5 sensors and 10% demand error, CUSUM finds all leaks the sensors can see at 4% of demand but 8% at 1% (3% of all leaks); demand error, not sensor noise or detector choice, sets the limit (simulated leaks; BattLeDIM L-Town was unreachable).
Reproducible operators and refereed trainingMeasured float32 reduction-order drift, RepOps-style canonical order, Merkle-committed training traces with a teacher-forced referee; per-step tolerance is a cumulative hidden-deviation budget.
How many honest runs does a slashing tolerance need?Calibrating slashing tolerances from honest runs: split-conformal false-slash law (n+1-k)/(n+1) is exact but only ~51–56% of calibration sets are ≤ α, PAC safety needs n ≥ ln δ/ln(1−α) (459 at 1%/1%), the margin is cheater hiding room (×1.26 at n=2000, Pareto-3), Gaussian plug-ins over-slash skewed drift up to 17×, pooled hardware calibration over-slashes the noisy class 4.8× and lets the quiet class hide 2.9×.
Should drift tolerances be per layer or on the total?Per-layer vs total drift tolerances: a tolerance per layer admits L·c(L,α)σ vs z_α√L σ on the total (6.6× at L=16), the sum test ignores how a cheat is spread while the max test catches 2.7σ in one layer but lets 16σ through spread over all, and a union test caps the hidden budget at the sum test's.

All 14 Detection & tolerances projects →

Other research topics

Proper scoring & elicitation28 Effort & peer prediction9 Verification games & slashing19 Detection & tolerances14 Markets as routers13 Prediction-market design9 Paying contributors8 Decentralised training systems23 Digital twins & sim-to-real164 LLM chip design loops40 Robot manufacturing loops53 Ambient AI assistant46 Self-replicating manufacturing34 Simulation-first design64 Engineered clothing45 Task-specific AI accelerators40 Algorithmic chip-design loops42 Open-source purpose-built robots44 The software-engineering loop43 Adversarial human imitation41 Open-source residential construction41 Autonomous multi-angle capture34 Sensing, imaging and 3D31 Energy supply and dense energy storage39 Robot societies among people31 Agentic co-design of low-cost robots27 Laundry-hanging robots25 Machine learning on qualitative data23 Robots on the building site27 Civilizations & group psychology1 History & human sciences1

Produced with AI assistance (Claude). Results are from the code in this repository and are simulated or stylised unless the abstract says otherwise; nothing here implies affiliation with or endorsement by any lab, researcher or company cited. MIT licensed.