Machine Learning
Research on Machine Learning: explanations, evidence, limitations and connections between papers.
96 results
Deadline and energy reshape time-series model choice
A handover-prediction study finds LightGBM used less measured energy than the best neural model under burst-schedule testing.
Random remasking lets diffusion models revise and improve sequences
Random Remask-Denoise lets masked diffusion models revisit guesses, improving reported logical-task accuracy as correction rounds increase.
Match changes in model depth to the cost and outcome you need
A language assistant may need a faster answer; a tabular predictor may need a smaller model. Adding a pass through shared weights increases computation without adding stored weights, while removing blocks changes the model itself. The same word, depth, hides different costs. These studies suggest choosing a change at the point where the cost arises, then judging it by the outcome actually tested. Their evidence covers training loss, downstream language tasks, word prediction and tabular classification under different conditions; it does not identify one best depth for every model.
Apple Silicon speedups depend on the workload and runtime
Tests of four LLM optimization families show Apple Silicon gains depend on workload, runtime, and hardware.
Graph prompts can collapse into nearly identical tokens
An audit finds graph tokens can collapse into generic prompts; a two-stage procedure reduces similarity, but task scores vary.
Mean proxies help sparse attention retain accuracy
OpenAttention uses cache proxies to keep mean accuracy near dense attention on seven long-context tasks at a 1% read.
An index maps which work AI agents could take over worldwide
The Agent Delegability Index estimates how much work AI agents could take over, with full handover rare in its global estimate.
Clinical model scores vary by decision type
Across public clinical tasks, model scores were strongest for knowledge and text, and lowest for volumetric findings.
Specify the event and clock before trusting a decision
A game planner can misprice one reward, a voice assistant can finish an outdated task, and a video monitor can alert at the wrong moment. An earnings-call benchmark can also start measuring a return before the input it describes is ready. Each case depends on where an event boundary is placed. The studies make those boundaries visible through transition substitution, task-version checks, alert rules, and declared price clocks. Their findings answer different questions under different protocols; the useful first step is to identify the event, the relevant state, and the clock before interpreting a result.
Changing model settings can reduce prediction coverage
An audit finds that changing inference settings can reduce conformal coverage on finite-label tasks, though effects vary by model and dataset.
When whole-task audits misjudge component checkers
A theory and ECG study show how whole-task audits can weaken component checkers, while retaining their failures preserved performance in tested settings.
A gate checks whether model updates really improve
BRIDGE checks model updates on separate data and reduces false improvements in tested tabular tasks; its population bound is synthetic.
When compilers can safely move MXFP4 scales
A compiler method proves when shared MXFP4 scales can move across a binary32 reduction without changing its result.
Agents can improve research code, but many tasks stump all
FrontierCS 2 tests agents improving research code; its top agent passed 58% of runs across 48 tasks under an all-metrics rule.
A satellite system brings stronger rainfall detection into real time
Huayu-Global combines three satellite streams and detects precipitation better overall than seven products in matched station comparisons.
For LLM inference, lower precision beats 2:4 above the floor
A comparison of 2:4 sparsity and quantization finds dense formats more accurate above the lowest supported precision in two 8B models.
A bit-based method compares context with model weights
An information-based method compares context with trained weights; issued certificates matched measured winners in a designed transformer test.
A flow-wide method improves low-temperature generation
Bayes-Consistent Tempering aligns training and sampling for flow language models, improving low-temperature results in tested math and Sudoku tasks.
A memory branch improves GraphCast’s longer-range forecasts
A memory correction branch lowered GraphCast’s weighted forecast loss across ten-day forecasts from 122 starts in 2023.
Match parallel generation to the work that slows it down
A phone agent can wait at every screen action, while a text generator can spend time proposing, checking, or reconstructing its output. Parallel work helps when it reaches the costly part of that particular workflow. Across these studies, moving or overlapping work produces reported throughput or latency gains in specific workloads and serving setups. JetSpec++ also reports a serving gain, while its predictions of future context remain estimates until verification. The results support judging each method by its local outcome, without forming a shared speed ranking.
Studies test final self-improving systems more than the loop behind them
A study of papers finds repeated tests of full improvement loops less common than evaluations of final systems.
A new method tracks how training targets change predictions
The target-response Jacobian closely tracks refitted prediction shifts after training-target changes on three tested tabular datasets.
A score can rank when CT systems fail without warning
A prospective score ranked silent failure in two learned CT systems, with evidence limited to simulated access tests.
Racer checks whether GPU kernel rewrites preserve their function
Racer symbolically checks GPU kernel rewrites and exposes bugs missed by sampled tests, within the coverage of its production-kernel study.