Advancing Model Research in AgentX: Long-Horizon Autonomy for Industrial Recommender Systems introduces AgentX-Model, a dual-agent framework that links proposal development with model experimentation in business-defined sandboxes. The Research Agent drafts proposals from literature and findings, while the Model Agent runs multi‑round experiments, returning code, metrics, and open questions. The framework cycles through Reproduce, Follow‑up, Composition, and Diagnose actions, achieving high AUC gains and significant business metric improvements in online A/B tests.
The paper introduces Evo-Rec, a three‑stage framework that improves generative recommendation by learning better reasoning traces for Semantic ID (SID) generation. First, it aligns SIDs with textual and behavioral contexts; second, it samples and selects reasoning traces that improve ground‑truth item prediction via supervised fine‑tuning; third, it refines the reasoning policy with reinforcement learning using catalog‑constrained generation and ranking‑aware feedback. Experiments on three Amazon Review datasets show Evo‑Rec consistently outperforms discriminative, generative, and other reasoning‑enhanced recommenders across all metrics.
ENDOPROMPT is a white‑box technique that learns prefixes to degrade the utility of language model responses without producing harmful content. It uses clean victim continuations as pseudo‑references, performing local search and preference fitting on unlabeled instructions to generate utility‑degrading prefixes. In experiments on four instruction‑tuned models and seven benign benchmarks, ENDOPROMPT reduces utility by an average of 26.8 percentage points, with 27 of 28 benchmark cells showing negative change.
The study compares Jev, a typed classifier that outputs probabilities over allowed answers, with three flash-tier LLM rubric judges across nine panels from seven benchmarks. Jev’s accuracy differs significantly from an LLM judge in only 8 of 27 paired comparisons, performing best on binary criteria and worse only on graded ones, while most other comparisons are inconclusive. In terms of cost and speed, Jev is 29 to 325 times cheaper and 30 to 220 times faster than the LLM judges, and a cascade that defers uncertain Jev verdicts to an LLM yields only modest gains of up to 2.0 points over the best single judge.
TopU-LBVS is a new multi‑target benchmark for ligand‑based virtual screening that addresses shortcomings of previous datasets by using hard‑negative decoys and a fixed 1:40 active‑to‑decoy ratio. It covers 93 protein targets across seven classes, provides three evaluation protocols (full, low‑data, and mini), and includes curated ChEMBL‑35 bioactivity data with property‑matched, structurally similar decoys to reduce shortcut learning. The benchmark comes with released data, fixed splits, evaluation code, and baseline implementations for reproducible comparison of LBVS and molecular representation methods.
The paper introduces Flow-Based Distribution Matching (FBDM), a non‑adversarial method that learns self‑supervised representations by aligning images to explicit geometric references through spherical conditional velocity regression. By using an ETF‑inspired reference, FBDM allows more reference components than the flow dimension while maintaining geometric separation, and it incorporates an alignment loss to bring augmented views closer together. Experiments on datasets from CIFAR to ImageNet show that FBDM performs nearly as well as adversarial distribution‑matching methods, achieves a 1.48‑ to 1.83‑fold speedup, and offers a theoretical bound on downstream misclassification rates.
The paper introduces m-WCN, an end‑to‑end deep learning framework that neuralizes multi‑wavelet decomposition to jointly extract temporal patterns and frequency components from time series. It enforces orthogonality constraints to produce interpretable multi‑resolution representations, and builds two task‑specific architectures—TFBC for classification and FTB for forecasting—on top of this foundation. Experiments on 64 UCR datasets and seven forecasting benchmarks show that TFBC and FTB outperform baseline models, achieving average improvements of about 20% in both classification and forecasting tasks.
PHOSA presents a photorealistic 3D sign avatar modeling framework that addresses the need for accurate communication with the Deaf community. The authors introduce MVSign, a multi‑view Chinese sign language dataset co‑designed with Deaf experts, and a hybrid fitting pipeline for precise SMPL‑X annotation. Their decoupled avatar representation and motion‑aware sampling achieve high‑fidelity visual results on MVSign and generalize to monocular sign language videos.
The paper introduces PixelJev, a native-image decision interface that combines an image, a task instruction, and a runtime candidate set to produce a structured choice and candidate-conditioned probabilities using small open multimodal models. It unifies recognition and multiple-choice visual question answering via a language-model readout, offering options for frozen inference, language-side adaptation, and held-out calibration. Across seven benchmarks, 64-shot source adaptation significantly boosts Pets accuracy from 60.13% to 92.40%, and the model supports VQA tasks without target fitting, though calibration and cross-family transfer remain challenges.
Spot, Separate, and Enhance (SSE) is the first multimodal, user‑guided generative model for audio remixing and enhancement. It rebalances audio, removes unwanted sources, and reduces reverberation in video content, guided by both video and textual descriptions. The authors introduce the DegradedMix dataset, built on MuddyMix, and use generative‑model evaluation metrics to demonstrate SSE’s superior controllability and remixing quality compared to existing baselines.
Repeated evaluation can estimate a benchmark score accurately while still requiring replication to certify narrow uncertainty. We characterize that requirement on a fixed grid of $M$ tasks with $L$ binary paths per task under the hard budget $(M+t)K$, where each path costs at most $K$ responses or episodes.
The paper introduces EASE, an encoder‑only model for Audio‑Visual Semantic Segmentation that eliminates redundant components found in prior Transformer‑based approaches. EASE achieves state‑of‑the‑art accuracy while running at up to 365 FPS—three times faster than previous models—and can be trained in under 11 GPU‑hours. The authors provide code, weights, and samples, positioning EASE as a scalable foundation for future research and real‑time applications.
RGBD20K is a new large-scale RGB‑D dataset designed to advance semantic segmentation research. It contains 20,000 image pairs annotated with 160 fine‑grained categories, far exceeding the diversity of existing benchmarks such as NYUv2 and SUN RGB‑D. The authors also provide high‑fidelity annotations and introduce a score‑purified fusion (SPF) method that achieves state‑of‑the‑art results on multiple benchmarks.
The paper investigates whether large language models (LLMs) assess politeness in ways that match human judgments. Using two English datasets—one with continuous ratings and another with three‑way categorical labels—the authors find that LLMs agree more with each other than with humans. They observe that model–human alignment depends on explicit linguistic cues, while misaligned cases often involve rapport‑building strategies. Additionally, models tend to overproduce Neutral labels and underpredict Impolite labels, a pattern that persists even when expert consensus is used as a reference.
The study benchmarks active spot selection methods against random sampling for spatial transcriptomics, focusing on cost‑efficient data acquisition. Using two public cohorts, the authors simulate multi‑round selection with uncertainty‑based (MC‑dropout, TOD) and diversity‑based (CoreSet, TypiClust) strategies, evaluating performance at 5%, 10%, 30%, and 50% of the spot pool. Results show that none of the active strategies consistently outperforms random sampling across all budgets or evaluation metrics, with performance varying by dataset and metric.
By Zheyu Zhu, Junchao Zhu, Fengbei Liu, Tianyuan Yao, Gelei Xu, John Cannon, Haichun Yang, Yuankai Huo, Mert R. Sabuncu, Ruining Deng
DeltaS is a query‑agnostic, training‑free method for evicting key‑value cache entries in hybrid video‑language models that combine linear and full attention. It uses the change in the recurrent state of gated‑delta linear attention—called state drift—to decide which video chunks to keep, selecting those that induce larger normalized state changes. In experiments with a fixed memory budget, DeltaS outperforms position‑, attention‑, and key‑value‑based eviction signals, improving performance by 2.1 points on average across six long‑video benchmarks and 5.6 points on the longest benchmark, while adding only 1.9% of the forward‑pass cost.
By Taeyoun Kwon, Seungjin Kim, Hyeonyu Kim, Moon Hwan Kim
The paper introduces Latent Action Driving Annotations (LADA), a three‑stage pipeline that converts large amounts of unlabelled observation‑trajectory data into a language‑conditioned driving model. First, a latent action model with a vector‑quantised bottleneck learns a compact codebook of vehicle intents. Then, a small set of language‑annotated examples trains a vision‑language translator to map observations and instructions into this codebook, and finally a VLA is trained on observation‑latent‑action pairs across the full corpus. Using less than 5% of language annotations, LADA attains a Driving Score of 87.98 and a Success Rate of 70.46% on Bench2Drive, matching or surpassing fully supervised baselines.
By Alexey Zakharov, Kemal Oksuz, Puneet K. Dokania
The paper presents a lightweight machine‑learning approach for multi‑class malware detection on resource‑constrained devices. Using a LightGBM classifier with SMOTE oversampling, SOM‑US undersampling, and Genetic‑Algorithm feature selection, the authors achieve 89.1 % accuracy on four malware families and 76 % on 16 individual malware types. A second Random‑Forest model further improves family classification to 91.2 % and individual classification to 78.7 %.
By Abdul Khalek Alve, Alif Rahman, Saadman Zaman, Sazzad Hossen Himel, Muhammad Iqbal Hossain
The paper presents a dual‑encoder Transformer model for estimating Planetary Boundary Layer Height (PBLH) from satellite radiances, addressing challenges of multimodal, spatially incomplete data. It benchmarks eight different approaches, analyzes model reliance via grouped Shapley decomposition, and demonstrates that the proposed architecture achieves a mean absolute error of 155.8 m on a global test set, outperforming all baselines. On out‑of‑distribution data from the TEAMx campaign, the model attains 165.3 m MAE, better than a pixel‑wise baseline trained on the same data.
By Lorenzo Innocenti, Luca Catalano, Edoardo Arnaudo, Claudio Rossi, Salvatore Larosa, Domenico Cimini, Paolo Garza
Fine‑tuning large language models on parallel data can improve translation quality but also causes catastrophic forgetting of general capabilities. The study evaluates several forgetting‑mitigation methods—anchored to auxiliary data, model outputs, and base model parameters—using Llama 3.2 1B Instruct and Llama 3.1 8B Instruct on Arabic‑English and Spanish‑English translation tasks. Elastic Weight Consolidation best preserves general benchmark performance, yet only data mixing with control‑task examples maintains instruction‑following abilities such as formality and grammatical gender control, though these gains do not generalize to unseen prompts.
By Niklas Scholz, David Thulke, Abdallah Nasir, Will Allred, Evgeny Matusov, Hermann Ney