Loading…
Topic
622 posts about Machine Learning. Every summary links to the original.
See the idea
A walkthrough of how a 3D CT model turns a voxel volume into a structured diagnostic report: isotropic resampling, 3D patch tokens, spatial encoding, and language-model reasoning.
The Good Stuff
Paid picks we use ourselves. Links carry a referral code — a subscription through one earns us a commission at no extra cost to you.
Newest first
TriNetX operates a federated real-world health data network spanning more than 230 healthcare organizations and needed infrastructure to make growing volumes of RWD easier to use for clinical research. Databricks now provides the centralized lakehouse architecture, consolidating electronic health record data while supporting custom datasets, consulting, ML models, proprietary algorithms, APIs and self-service access. TriNetX is extending this foundation with a beta Query Assistant for natural-language analysis and a Support Assistant prototype built with Databricks’ Agent Bricks, requiring no programming expertise for the former. In 2025, pharmaceutical clients reduced protocol amendments by up to 50%, while an AI-enhanced site-identification effort achieved a 63% site acceptance rate and nine-day average response time. Other models suggest enrollment conversion could rise from 33% to 85%, with expanded APIs and Genie exploration planned for 2026.
AIOps applies AI and machine learning to IT operations to analyze logs, metrics, traces, events, alerts, and related system data. It detects anomalies, correlates events, uses topology and dependencies to identify likely root causes, and supports guided or automated responses. The guide positions AIOps between observability and action, reducing noise and adding context without replacing observability, DevOps, or human judgment. Its workflow includes data ingestion, normalization, anomaly detection, event correlation, root-cause analysis, automation and orchestration, and collaboration workflows, with human-in-the-loop review recommended for higher-risk production changes. It contrasts domain-centric and domain-agnostic approaches and recommends evaluating coverage, depth, integrations, setup effort, and organizational fit before expanding automation.
Airbnb’s guest sequence recommender previously refreshed Transformer-based embeddings in a daily batch, leaving activity from earlier the same day unavailable until the following run and creating up to nearly two days of staleness. To close this gap, Airbnb added Push Mode, which emits notifications when streaming feature values commit, and Near-real-time Model Transform, which invokes the deployed sequence encoder inside that streaming pipeline. Each page view or search is merged with the guest’s long-term and short-term sequence data, triggers inference, and writes a fresh embedding to the same low-latency store used by batch serving. The resulting system typically updates embeddings in under a minute, often within 10–30 seconds, improved offline NDCG by 1.67%, and increased uncancelled bookings by approximately one-third of a percent in online A/B tests.
JetBrains Context is presented as a production RAG pipeline for semantic code search, designed to give coding agents precise, citable evidence from repositories instead of relying on keyword tools such as grep. This first installment covers parsing, structure-aware chunking, and vectorization, with the goal of retrieving meaningful functions, symbols, or snippets rather than whole files or insignificant lines. Language parsers produce syntax nodes, and the algorithm retains the largest units that fit, splits oversized leaf nodes when necessary, groups small adjacent units, and keeps documentation, annotations, modifiers, and closing syntax with their constructs. JetBrains Context supports this approach for Kotlin, Java, Python, JavaScript, TypeScript, C#, PHP, Go, and Rust; other languages use line-based splitting. Vectors are generated with an open-weight model on company-operated GPUs, while indexed records store coordinates rather than source code, and the post reports that this setup outperformed hosted embedding APIs on internal code-retrieval benchmarks.
REVERSAL-BENCH addresses the challenge of evaluating reset-free reinforcement learning when environments contain irreversible transitions. It controls reversibility with continuous parameter ρ ∈ [0, 1] and supplies a reset oracle, a ground-truth mechanism for verifying state recoverability across eight manipulation settings and five physics engines. Evaluations spanning standard actor-critic, safe reinforcement learning, and specialized reset-free methods find a sharp reversibility cliff: reset-free agents become absorbed into irrecoverable states as ρ increases, while episodic agents maintain steady learning. Geometrically identical reversible counterparts indicate the breakdown is caused by irreversibility rather than obstacle complexity, and the effect persists under learned policies in full physics simulations. The released suite includes a multi-simulator dataset with recoverability labels and a reset oracle; a safety shield predicts recoverability accurately, but active recovery mainly works when agents can steer away from traps.
Modern storefronts can appear healthy while malicious JavaScript diverts affiliate credit, hijacks searches or clicks, tampers with analytics, or loads further code. Cloudflare describes four operations spanning eight payloads that Page Shield ML detected in live traffic; humans verified the findings afterward, while seven payloads were absent from VirusTotal and none received a malicious verdict from URLScan. The detection system represents JavaScript as a graph with a graph neural network (GNN), sends fewer than 0.3% of analyzed scripts to a lightweight LLM on Workers AI, and uses a teacher cohort of frontier models for difficult investigations. This combination handles minification, renaming, some obfuscation, and behavior gated by device, time, geography, referrer, browser state, or cooldowns. The post concludes that continuous browser visibility and runtime context are needed because static, one-time scans can miss selectively executed attacks.
Discrete flow matching can parallelize text generation across positions, but reaching high quality may require hundreds or thousands of model evaluations. The paper argues that few-step distillation fails primarily because teacher trajectories are built from blind stochastic jumps: an early poor midpoint decision propagates, and the student is trained to imitate it. Its Trajectory-Shaped Discrete Flow Matching (TS-DFM) uses a lightweight energy compass during training to score candidate continuations at each midpoint and select the most coherent path, without adding inference cost. In experiments with 170M-parameter language models, an eight-step shaped student achieves 32% lower perplexity than a 1,024-step teacher and runs 128 times faster, with gains across source distributions and three evaluators; it also outperforms the compared discrete-generation baselines, including systems using more data or larger models.
Airbnb describes Insight Miner, an agent harness designed to make unstructured data investigations scalable, reproducible, and audit-ready. Its problem was a bespoke process spanning data discovery, expert review, representative datasets, and manually curated methods that could not keep pace with new languages, geographies, and LLM-based products. The shared package combines extraction, embedding, and clustering with prompt tuning, hard-example mining, contrastive labeling, and a path from unsupervised exploration to classification, while an agent handles research questions, execution, and methodological guidance. Airbnb says investigations that once took months can now take days, allowing mechanical work to scale while data scientists spend more time inspecting ambiguous evidence, testing hypotheses, and improving techniques. The tool also lets nontechnical experts run analyses across use cases including survey coding, model evaluation, and fraud patterns.
Together AI is expanding its Fine-Tuning service to cover newer open-weight models and more of the workflow from dataset inspection through model selection. The update adds live training and evaluation metrics through the API, CLI, dashboard, and Python SDK, including curves that can be compared across in-flight jobs and underlying series that can be exported. Training controls now include Expert LoRA for mixture-of-experts layers, early stopping based on validation loss, and arbitrary effective batch sizes through gradient accumulation. The post reports that Expert LoRA recalled up to 89% of 200 invented facts versus 15% for attention-only adapters, while listed LoRA training prices fall by 30% to 70% for selected models, including the gpt-oss series.
Repeatedly optimizing against long-lived machine-learning benchmarks appears to invite overfitting, yet the post notes that gains on heavily reused benchmarks have largely transferred to newly created test sets. It describes a paper that resets LLM-based research agents, gives an explorer repeated access to validation feedback, and passes the resulting strategy through a token bottleneck to a fresh reproducer. Across eight datasets, strategies from hundreds of experiments were reproducible from prompts as short as 16 to 32 tokens; at shorter budgets, performance diverged, showing that the compressed message carried data-dependent choices rather than only prior knowledge. Deliberate overfitting occurred in 38 of 102 runs, with validation accuracy more than 10% above held-out accuracy, but those gains disappeared after compression; possible pretraining side channels remain unresolved.
Anthropic, an AI safety and research company, raised $580 million in Series B financing to build large-scale experimental infrastructure for studying and improving the safety properties of computationally intensive AI models. Since its 2021 founding, the company has researched steerability, robustness, and interpretability, including mathematically reverse engineering small language models and investigating pattern-matching behavior in larger ones. It developed baseline techniques and reinforcement learning methods for making models more “helpful and harmless,” released an alignment-focused dataset, and analyzed sudden performance changes and their societal impacts. The stated goal is to create models with stronger implicit safeguards, fewer after-training interventions, and tools for examining whether those safeguards work, alongside policy and societal research. The round was led by FTX CEO Sam Bankman-Fried and followed a $124 million Series A in 2021; Anthropic said its team numbered around 40 people.
Anthropic, an AI safety and research company led by CEO Dario Amodei and President Daniela Amodei, has raised $124 million in a Series A financing round. The funding will support its research roadmap, computationally intensive work, and prototypes of reliable and steerable AI systems, with an emphasis on systems that are interpretable and robust. Its immediate priorities include increasing the reliability of large-scale AI models, developing techniques and tools for interpretability, and integrating human feedback more tightly into AI development and deployment. The round was led by Jaan Tallinn, co-founder of Skype, and included James McClave, Dustin Moskovitz, the Center for Emerging Risk Research, Eric Schmidt, and others; Anthropic is hiring researchers, engineers, and operational experts.
IBM has released Granite Time Series PatchTST-FM-r2, a roughly 385-million-parameter foundation model for zero-shot forecasting across demand, prices, energy loads, traffic, telemetry, and other time series. The model uses conformer blocks combining multi-head self-attention with temporal convolution, 50% overlapping patches with Hamming-window weighting, overlap-and-add forecasting, contexts up to 8,192 steps, and a 99-quantile prediction head for probabilistic outputs and imputation support. On GIFT-Eval, it ranks second for both CRPS and MASE among replicable zero-shot models as of September 8, 2026, and first in that category among permissively licensed models; including pretrained models, it ranks third for CRPS and fourth for MASE. IBM provides open weights, architecture, inference code, and reproducibility code under Apache 2.0 and OpenMDW 1.0. The release also describes Python use and streaming integration through Apache Flink on Confluent Cloud.
Autoregressive LLM inference is memory-bandwidth bound because each token requires a full pass over billions of parameters, while speculative decoding can verify several drafted tokens in one pass. Speculators 0.6.0 adapts a single native MTP head with FastMTP-style recursive fine-tuning, addressing the mismatch between teacher-forced training and vLLM’s autoregressive reuse of the head. The training loop fuses each next-token embedding with the current hidden state, predicts the following token, feeds the resulting hidden state back, and weights three steps with normalized exponential decay using β = 0.6. Native mtp.* weights extraction, shared full-vocabulary embeddings and output head, and vLLM-compatible stitching support deployment; on Qwen3-Next-80B-A3B-Instruct evaluated on GSM8K, mean accepted length improved at positions 0–2 from 0.897, 0.719, 0.476 to 0.912, 0.776, 0.616.
Botika builds agentic e-commerce teams that automate visual production for global fashion brands, using proprietary foundation models and a 100-terabyte image dataset. The company runs its entire full-stack AI operation on Modal, with the data pipeline processing 100 terabytes at less than 1% error rate across thousands of containers. Before Modal, the team struggled with Kubernetes fleet management and slow GCP Batch jobs; now a single researcher can launch 50 experiments daily using Claude, and they set up reinforcement learning infrastructure in less than a day. Modal handles autoscaling, cold starts, and multi-node training with RDMA, allowing Botika to double traffic in seconds, and they serve about 15 models on L4s, L40S, A100s, and H100s. The infrastructure is not Modal-specific, enabling the team to run open-source and custom frameworks without modification.
REFACTOR-VLA addresses the weakness of monolithic vision-language-action (VLA) models, which generate raw commands or short action sequences and struggle with long-horizon tasks and interpretability. Its wake/sleep architecture uses a Behavioral-Equivalence Kernel (BEK) to cluster motor-program segments by outcomes from a learned latent world model, then produces typed lambda terms for a library-conditioned rectified-flow action decoder. Skills are admitted only when they satisfy both Minimum Description Length (MDL) and return-preservation gates, while training proceeds through world-model warmup, wake-phase policy optimization, and sleep-phase skill discovery. On the full LIBERO benchmark, expanding the world model from 188 million to 430 million parameters worsened results on all four suites, showing that scale alone was not beneficial. Adding supervised contrastive InfoNCE loss during warmup substantially improved sleep-phase clustering, with reported NMI scores of 0.462, 0.867, 0.915, and 0.754 for the Object, Spatial, Goal, and LIBERO-10 suites, respectively, under n = 3 multi-seeding.
Autonomous-driving perception changes when the same software moves across carlines because sensor placement, calibration, fields of view, occlusions, geometry, timing, and coverage change. The post presents NVIDIA Omniverse NuRec as a way to adapt a stack before a complete target-carline dataset exists, while retaining real-world data for grounding and validation. NuRec reconstructs recorded drives with 3D Gaussian splatting, renders novel views through a target rig with configurable camera models and calibration, and preserves trajectories, annotations, and map data for label adaptation. The workflow pairs a reconstructed scene with a rig, renders target views, refines frames with NVIDIA Harmonizer, and trains a perception model on the output. It also documents containerized rendering and auxiliary-data generation, including semantic masks, metric depth, and ego-vehicle masks, while noting that views outside observed coverage may be lower quality.
This practical comparison examines how YOLO12, YOLO26, and RF-DETR behave when moved from COCO to specialized real-world object-detection datasets, and how fine-tuning changes the outcome. The authors first evaluate six pretrained checkpoints on COCO val2017, then use cable-damage, bone-fracture, and soda-bottles datasets from RF100-VL for zero-shot testing, fine-tuning, and accuracy/latency comparison. Because the model implementations require incompatible dependencies, the workflow uses one PyCharm project with three isolated uv environments and remote GPU execution; the reported latency stays in native frameworks rather than TensorRT. All models perform well on COCO, but direct performance on specialized data falls close to zero, while 10 fine-tuning epochs adapt the detectors well to cable damage and soda bottles. Bone-fracture remains difficult despite fine-tuning, with RF-DETR performing best, indicating that domain shift may require more data, longer training, or domain-specific pretraining.
Cloudflare introduces Adaptive Intelligence, a bot detection engine designed for attackers who distribute requests across proxies, vary fingerprints, and continually test defenses. It argues that deterministic, rule-based systems create stationary targets: attackers receive repeated yes-or-no feedback while defenders update on slower release cycles, making adaptation economically cheaper for attackers. The engine changes this dynamic through continuously retrained machine learning, statistical rather than fixed judgments, and planned disposable rules that appear and retire at random intervals; it also learns from corrections and misses across millions of sites. Its first component, continuous retraining of the machine-learning system behind bot score, is launching now, while disposable rule generation and broader traffic learning are described as forthcoming, with Enterprise customers enabling “Auto Update Machine Learning” to receive it.
Construction data is fragmented across systems, organizations, equipment types, and formats, limiting the operational context available for decision-making. Trackunit addresses this data-intelligence problem with IrisX, an operating data platform built on Databricks that connects equipment, machine, operator, site, and operational data. Its three-part model—connect, distill, and amplify—combines data engineering, analytics, and AI on an open platform, then delivers governed, construction-specific intelligence through applications, workflows, and AI interfaces. The platform supports natural-language analysis of equipment signals and contextual factors such as maintenance history, operating conditions, location, contracts, project needs, and asset availability. Examples described include battery-management guidance for OEMs, missed-invoice detection for rental companies, and asset redeployment for contractors, with reported value of approximately $3 million and $2 million in two examples.
The best engineering reads, once a week.
