Search

The article titled "Bibliometric Analysis and Co-word Mapping: The Field of Knowledge Graphs" organizes and reviews literature on the knowledge graph research field using bibliometric and co-word analysis, with references spanning knowledge graph definitions and surveys, embedding methods, completion and refinement, domain-specific graphs, educational applications, and scientometric and co-word methods themselves; however, the text available here is only the reference list, and the body, figures, and specific bibliometric results are not included.

This qualitative phenomenological study interviewed 10 virtual elementary teachers of kindergarten through fifth grade at a large elementary virtual school and used Creswell and Poth's (2018) data analysis spiral to derive four themes—efficiency, preservation of the teacher role, instructional enhancement, and AI literacy—showing that teachers simultaneously weigh efficiency and enhanced instructional quality against misinformation, privacy protection, ethical concerns, and misuse, and view generative AI as a dynamic tool whose value depends on thoughtful, informed, and human-centered application.

Using 1312 photographs of infant stool in diapers (1100 meeting quality criteria, of which 896 had majority agreement among three pediatric gastroenterologists as the reference set), the study trained 20 convolutional neural network models on Teachable Machine with an 85/15 training and internal-validation split and assessed agreement with specialists on the BITSS: the best seven-category model reached overall accuracy 0.56 with linear and quadratic weighted Kappa of 0.59 and 0.75, while a four-category grouping (constipated, formed, soft, liquid) reached accuracy 0.76 with weighted Kappa of 0.69 and 0.78, whereas agreement among the three specialists showed a Fleiss Kappa of only 0.24 and 24.7% complete agreement.

Google announced Gemini 4 Argon, a frontier model first rolling out to trusted cyber defenders through its Fairwind Program, built for long-horizon deep reasoning, raising the output token limit from 64K to 1M and reporting 77.9% on DeepSWE v1.1, 91.7% on LVBench, 68% on CWE-bench v1, and 51.3% on AutomationBench, alongside internal cases including a 40% improvement over a published baseline in quantum subroutine optimization, over 300 TiB of memory freed in data centers, and a Rust port of libgav1 that replaced 32K lines of SIMD code and runs 2.7x faster than the prior Rust port with identical video output.

NVIDIA announced it is expanding xio-sig to include cuObject alongside cuFile and is making the cuObject client and server libraries generally available, while introducing a SCADA Server SDK so AI accelerators can access file and object storage over RDMA without routing data through the server CPU, with IBM demonstrating interoperability through a prototype integrating SCADA and IBM Storage Scale.

IEEE announced the recipients of its 2027 Technical Field Awards, selected by the IEEE Awards Board on the basis of technical achievement, societal impact, and leadership and approved by the IEEE Board of Directors, with recognized work spanning clinical translation of therapeutic ultrasound, heat-assisted magnetic recording, nonlinear and robust control, microwave and millimeter-wave circuits, speech and speaker recognition, multidimensional signal processing, MOS device physics, mobile communication, Internet traffic analysis, power electronics converters, measurement and standardization of data converters, integrated-circuit testing, image processing and computer vision education, data-center networking, the HEVC international standard, programmable digital signal processing, CMOS proce

NVIDIA announced that its 26th Graduate Fellowship Program is accepting worldwide applications for the 2027–2028 academic year, offering doctoral students in AI, machine learning, autonomous vehicles, computer graphics, robotics, healthcare, high-performance computing and related fields up to $60,000 plus mentors and technical support, with an application deadline of October 30, 2026 and a mandatory in-person internship at an NVIDIA research office in summer 2027 before the fellowship year.

Using Claude to score roughly 19,000 O*NET task descriptions across about 900 occupations by how controlled an environment a robot needs, the authors build a robot exposure index and find that today's robots can perform 74% of US physical tasks (34% of working hours) but are cost-competitive for only 0.3% of work, while a backtest since 1977 shows jobs with higher exposure later saw wage and employment declines.

The authors propose RegLLM, a diagnostic harness for bounded autonomy in regulated agentic workflows that instruments six trustworthiness signals (citation validity, source grounding, schema compliance, escalation correctness, constitutional alignment, unsafe-action rate) alongside a deterministic runtime supervisor; an offline reference run (n=12) lifts escalation recall from 0 to 0.67 and cuts the unsafe-action rate from 0.33 to 0.08, while two nominally identical single-GPU Qwen2.5-3B LoRA/DPO pilots (n=8, same seed and eval split) show task success of 0.25 versus 0.12 and escalation recall of 1.0 versus 0.

The work defines the length-scaling tax (LST) as the excess response length that RL post-training adds on already-solved queries without a matching accuracy gain, and proposes Length Self-Distillation (LSD), which routes solved prompt groups to on-policy distillation while keeping the original RLVR objective for unsolved groups, using an exponential moving average of the same policy lineage as teacher with no external model; LSD reduces LST from 19.0% to -3.7% on single-turn reasoning and from 31.4% to 13.7% on multi-turn agentic tasks while matching or improving average Pass@1 over RL.

DAGent introduces Evaluate-then-Grow incremental planning, in which an Orchestrator grows the task DAG one batch at a time conditioned on confidence and uncertainty signals from completed nodes, paired with a hierarchical context layer that propagates compact QueryDocs by default while preserving full execution traces for on-demand recall, and with DAGRPO, a DAG-conditioned RL adaptation combining topology-conditioned credit with structural compliance regularization; it surpasses the strongest open-source baseline by 5.3/5.8/2.0 points on BrowseComp-Plus, GAIA and xbench-DeepSearch at the Qwen3-235B-A22B scale, improves over a same-budget outcome-only GRPO baseline by 3.

The work introduces label-free bias-only test-time reinforcement learning: the pretrained backbone stays frozen while only about 100K bias parameters (down_proj.bias across 28 MLP layers) are optimized against majority-vote pseudo-label rewards, reaching 76.67% on MATH-500 with Qwen2.5-7B and 79.50% with Qwen2.5-Math-7B, improving vision-language and audio reasoning under the same procedure, and transferring frozen bias vectors to 4,500 verified-disjoint MATH problems to lift Qwen2.5-7B from 46.3% to 70.9% and Qwen2.5-Math-7B from 52.5% to 75.4%, with pseudo-label reliability and accessible gradient energy explaining why a restricted subspace still adapts.