• First graphon-level transfer without node correspondence — aligns graphs via Gromov–Wasserstein and transfers edge structure nonparametrically.
• Residual smoothing unlocks small/sparse targets with convergence & stability guarantees; SOTA on link prediction and graph classification.

• New minimax theory for "near-manifold" shift: exposes a sharp phase transition controlled by the support gap between target and source neighborhoods — unifying multiple geometric-transfer regimes.
• Ratio-free, adaptive estimator: achieves near-optimal, dimension-adaptive rates without density ratios and without assuming known geometry (works under approximate manifold mismatch).

• Bridges the text–time-series modality gap: introduces a Temporal Evolution Semantic Space (TESS) that distills free-form text into interpretable temporal primitives (mean shift, volatility, shape, lag), instead of directly fusing noisy token embeddings.
• LLM-guided yet numerically grounded forecasting: uses structured prompting + confidence-aware gating to inject reliable semantic signals as prefix tokens into a Transformer forecaster, yielding robust gains under event-driven non-stationarity (up to 29% error reduction).

• Sinkhorn entropic-OT coupling enables many-to-many region alignment across cities — no node matching required.
• OT-weighted contrastive loss + target-aware prototype hub prevents collapse and scales cleanly to multi-source heterogeneity.

• Inconsistency-aware multimodal ranking: studies how cross-modal discrepancies may provide complementary ranking evidence or degrade fusion quality — explicitly modeled in training.
• Cross-modal hard negatives + synergy-aware ranking loss: proposes CHNS to mine modality-specific hard negatives across branches, and a Synergy-aware BPR loss to ensure the fused branch achieves stronger preference margins than unimodal branches.

• Hybrid spectral–spatial modeling for domain shift: integrates a Mamba-based global spectral encoder with CNN local feature extraction, capturing long-range dependencies while preserving fine-grained spatial structure.
• Dual-level transfer via distillation + graph alignment: performs teacher–student knowledge distillation for distribution alignment and OT-guided graph consistency across domains, yielding robust cross-domain generalization under severe spectral mismatch.

• SSGP prunes dense scientific graphs into task-adaptive subgraphs via rank-based ensemble scoring — drastically shrinks agent search space.
• Reuse–patch execution + confidence-weighted aggregation boosts reproducibility, stability, and success rate of LLM scientific agents.

A multimodal AI app that reads text and images — captions, poems, screenshots, moodboards, all those tiny digital traces of self-expression — and whispers back the MBTI vibe it gives off. It doesn't claim to tell you who someone really is. It asks a softer, kinder question instead.

For anyone who has ever left a movie wondering — what if it ended differently? Insert a new scene, change one choice, shift a single moment. It stays close to the emotional soul of the original — preserving its tone, longing, and ache — while imagining endings that feel tender, devastating, hopeful, or quietly healing. Not every ending needs to be undone. But some deserve to be imagined differently.

Share what is on your heart, and receive a letter from the movie or TV character who would understand. More love letter than chatbot — intimate, tender, just a little magical. The comfort lies in its emotional closeness: it offers not advice, but presence. The rare feeling that a voice from another story has stepped out of the screen to sit beside you for a while.

For anyone who has ever wanted to slip away for a day or two. Instead of a typical itinerary, it builds a small borrowed life in another city based on your mood — cinematic, soothing, touched with a bit of humor. Not because it solves anything, but because it gives you a brief imaginative place to rest, wander, and feel held by another version of life. Scenes, snacks, photographs, and the kind of inner weather that only changes when you leave home.

Step into a curated archive of parallel selves. Instead of planning a trip, it generates a small, emotionally resonant keepsake — a ticket, a postcard, a receipt, a note — from a life you've been quietly standing outside of. Cinematic, intimate, collectible. An invitation to imagine not just another city, but another self you might still be growing toward.

A map-based cultural discovery app for collecting meaningful places and exploring what lives around them beyond geography alone. After adding a city, you move from the global atlas into a curated layer of screen references, books, and local landmarks — so a place becomes not just somewhere on the map, but a small world of its own. Editorial, archive-inspired, cultural browsing as cartography.

An atmospheric AI literary experience inspired by García Márquez's rain-soaked world of memory, fire, butterflies, and magical realism. Rather than retelling the novel, it invites you into a room, a ritual, and a fate of your own. A small sequence of questions transforms mood into a story fragment that feels less like generated text and more like something recovered from an old archive in Macondo itself. Dark-gold palette, burning-paper visuals, haunted editorial layout.