Unverified50% confidenceFactExact time
The two-stage system (Deep SVDD + narrow CPU classifier) increases the proportion of inputs that are "correctly classified or correctly rejected" by 15.0–19.4 percentage points, a significant improvement over classifiers without a rejection mechanism.
1
Sources
50%
Confidence
Long-term
Relevance
11/3/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedAt the retrospectively selected peak test accuracy for each method, the estimated energy consumption per input is 1.93–6.99 mJ for the Deep SVDD two-stage system, 52.53 mJ for Monte Carlo Dropout, and 8.76–17.51 mJ for deep ensembles.72% similarUnverifiedOn 32 SV-COMP tasks, raw verifier feedback raised the average correct model determinations from 12.7 to 22.0, but at the same time increased incorrect determinations from 1 to 7.70% similarUnverifiedCompared to the previous best system, MLR delivers nearly a fivefold improvement in identifying critical errors (from 14.81% to 73.43%).67% similarUnverifiedA learnable permutation-invariant graph context aggregation mechanism (implemented with weighted averaging and Set2Set) achieved the largest improvement, about 0.9 percentage points, on the CVRP100 to CVRP20 transfer.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1129616API
curl https://kongchang.com/api/v1/knowledge/claims/1129616MCP
get_claim(id=1129616)