Showing papers from Boston University Show all papers
Atlanta Poster Session 3, Thu, Dec 10, 10:00 AM–1:00 PM, Hall C1 · Published 2026
Readers and the AI panel: vote on this paper to see what they said.
Only vote on papers you've read. Sign in with GitHub to vote.
Paris Poster Session 2, Wed, Dec 9, 5:00 PM–7:00 PM, Paris Poster Hall · Published 2026
Readers and the AI panel: vote on this paper to see what they said.
Only vote on papers you've read. Sign in with GitHub to vote.
Atlanta Poster Session 5, Fri, Dec 11, 10:00 AM–1:00 PM, Hall C1 · Published 2026
Readers and the AI panel: vote on this paper to see what they said.
Only vote on papers you've read. Sign in with GitHub to vote.
Atlanta Poster Session 1, Wed, Dec 9, 10:00 AM–1:00 PM, Hall C1 · Published 2026
Readers and the AI panel: vote on this paper to see what they said.
Only vote on papers you've read. Sign in with GitHub to vote.
Protein folding models share a two-stage trunk mechanism initializing biochemical signals then spatial features, with causally steerable, interchangeable representations across architectures.
Sydney Poster Session 6, Thu, Dec 10, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026
Readers and the AI panel: vote on this paper to see what they said.
Only vote on papers you've read. Sign in with GitHub to vote.
Swift Sampling uses Taylor-series projections of visual feature trajectories to select temporally surprising frames, cutting overhead by 30x while boosting long-video accuracy up to 12.5 points.
Atlanta Poster Session 5, Fri, Dec 11, 10:00 AM–1:00 PM, Hall C1 · Published 2026 · ▲ 7 on Hugging Face
Readers and the AI panel: vote on this paper to see what they said.
Only vote on papers you've read. Sign in with GitHub to vote.
Economic model combining scaling laws with microeconomics shows profit-optimal LLM training scales near-linearly with hardware efficiency and sub-quadratically in cost, while current expenditure trends are only optimal under compute-bound assumptions.
Sydney Poster Session 2, Tue, Dec 8, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026
Readers and the AI panel: vote on this paper to see what they said.
Only vote on papers you've read. Sign in with GitHub to vote.
Regularized Newton training of overparameterized neural networks converges to a deterministic NNTK limit with exponentially fast uniform convergence across all frequencies, avoiding gradient descent's spectral bias.
Sydney Poster Session 6, Thu, Dec 10, 5:00 PM–8:00 PM, Hall 1-4 · Published 2026
Readers and the AI panel: vote on this paper to see what they said.
Only vote on papers you've read. Sign in with GitHub to vote.
A unified framework bounds DP privacy leakage against multi-target membership, attribute, and reconstruction attacks using only privacy parameters and adversarial baseline success rates.
Sydney Poster Session 3, Wed, Dec 9, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026
Readers and the AI panel: vote on this paper to see what they said.
Only vote on papers you've read. Sign in with GitHub to vote.
GenScale benchmarks relative object scale in image generation and editing, finding current models unreliable, while Rescale improves scale plausibility via localized correction.
Atlanta Poster Session 6, Fri, Dec 11, 4:30 PM–7:30 PM, Hall C1 · Published 2026
Readers and the AI panel: vote on this paper to see what they said.
Only vote on papers you've read. Sign in with GitHub to vote.
Activation patching's natural indirect effect embeds hidden interaction effects between components, which cause conditional importance to be invisible or inflated, explain faithfulness instability, scale with activation distance, and diagnose when greedy component ranking misses combinatorial mechan
Sydney Poster Session 1, Tue, Dec 8, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026
Readers and the AI panel: vote on this paper to see what they said.
Only vote on papers you've read. Sign in with GitHub to vote.
PAIR reframes diffusion model concept erasure via unsafe-safe pairs, using paired semantic realignment and directional Fisher-weighted adaptation to remove targeted concepts while preserving structural and semantic consistency.
Sydney Poster Session 5, Thu, Dec 10, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026
Readers and the AI panel: vote on this paper to see what they said.
Only vote on papers you've read. Sign in with GitHub to vote.
ReToken introduces one learnable retrieval token that selects sparse visual tokens from long contexts, improving vision-language models by up to 13.4 points on visual retrieval while fitting on a single GPU.
Atlanta Poster Session 6, Fri, Dec 11, 4:30 PM–7:30 PM, Hall C1 · Published 2026 · ▲ 8 on Hugging Face · Code ★ 18
Readers and the AI panel: vote on this paper to see what they said.
Only vote on papers you've read. Sign in with GitHub to vote.
ACC++ improves circuit tracing to extract interpretable prompt-specific language model circuits from single passes, revealing clustered indirect-object mechanisms and language-specific reused components.
Atlanta Poster Session 1, Wed, Dec 9, 10:00 AM–1:00 PM, Hall C1 · Published 2026
Readers and the AI panel: vote on this paper to see what they said.
Only vote on papers you've read. Sign in with GitHub to vote.