The upcoming meeting, taking place on November 10 at Mila, will explore how we can collectively develop, govern, and deploy high-performing, reliable, and secure agentic systems by connecting academic researchers, industry experts, and practitioners.
Join us on the Venture Scientist Bootcamp, a full time, 4-month incubator at Mila, built specifically for deep tech founders with elite STEM backgrounds. You have until November 21, 2026 to apply!
We use cookies to analyze the browsing and usage of our website and to personalize your experience. You can disable these technologies at any time, but this may limit certain functionalities of the site. Read our Privacy Policy for more information.
Setting cookies
You can enable and disable the types of cookies you wish to accept. However certain choices you make could affect the services offered on our sites (e.g. suggestions, personalised ads, etc.).
Essential cookies
These cookies are necessary for the operation of the site and cannot be deactivated. (Still active)
Analytics cookies
Do you accept the use of cookies to measure the audience of our sites?
Multimedia Player
Do you accept the use of cookies to display and allow you to watch the video content hosted by our partners (YouTube, etc.)?
SpecAugment is the dominant speech augmentation policy, but researchers have studied it primarily for automatic speech recognition (ASR), an… (see more)d it operates on spectrograms, limiting its applicability to other speech tasks and waveform models. We propose Wav2Aug, a drop-in waveform augmentation policy that randomly samples two augmentations from a calibrated pool of waveform transformations. We calibrate the augmentation hyperparameters once and reuse them across tasks and architectures, eliminating task-specific tuning. Experiments on ten speech benchmarks show that Wav2Aug achieves ASR performance comparable to SpecAugment while improving over it on average across non-ASR tasks, yielding an average relative error reduction of 12.8% over the strongest compatible SpecAugment baseline. We further show that multi-view training provides additional gains, with two augmented views offering the best tradeoff between performance and computational cost.
Human-level dexterity remains a central open problem in robotics, yet few simulation frameworks cover the full workflow of dexterous manipul… (see more)ation, from setting up a robot to collecting dexterous demonstrations and evaluating policies. We present DexSuite, an all-in-one simulation framework for dexterous manipulation built on the Genesis simulator. DexSuite provides 30 tasks and assembles its 10 arms and 12 end-effectors, from parallel-jaw grippers to dexterous hands, into more than 100 single-arm robots and thousands of bimanual pairs. It also ships a standardized environment API, integrated teleoperation, and simulated tactile sensing. Moreover, we define a benchmark of 16 task-embodiment pairs spanning diverse contact complexities and horizons, and release a dataset of 4,700 successful episodes in simulation from human teleoperation, scripted experts, and learned policies. We release DexSuite as open source and hope others will build on it. Videos and datasets are available at https://dexsuiteorg.github.io
Interpersonal physiological synchrony is well documented in close relationships, where it supports social bonding and emotion regulation; ho… (see more)wever, little is known about whether cardiac synchrony emerges during emotional interactions between strangers. Here we examined interpersonal cardiac synchrony in pairs of strangers during an emotional sharing task. Eighty participants (40 dyads) engaged in 15 min of emotional sharing, during which one participant recounted a negative personal experience while the other provided support. We assessed cardiac synchrony as the cross-correlation of residualized inter-beat interval (IBI) time series from synchronized electrocardiogram (ECG) recordings. Results showed that greater mutual liking, gender match, and shared emotional perception were linked to stronger cardiac synchrony, suggesting that interpersonal synchrony reflects social and emotional alignment even between strangers. Additionally, the similarity between the heart rate variability (HRV) of each interaction partner predicted greater synchrony, indicating that physiological compatibility may facilitate alignment. We also examined whether a brief mindfulness induction given to the support provider enhanced synchrony and interaction quality. However, the mindfulness induction had no significant effect, suggesting that a brief self-regulation practice may not influence physiological alignment in first-time interactions. These findings extend research on interpersonal cardiac synchrony beyond established relationships, identifying key factors that shape physiological alignment during emotional sharing in first-time interactions.
Radiomics extracts quantitative features describing tumor shape, intensity, and texture from medical images to build models that predict cli… (see more)nical outcomes. However, large feature sets increase the risk of overfitting in small datasets, and building an end-to-end workflow often requires programming expertise. We present MEDiml (Medical Image Analysis with Machine Learning), open-source radiomics software available as a Python package (MEDiml-pkg) and a no-code desktop application (MEDiml-app). Both run on the same computational engine, and code generation exports graphically configured workflows as executable Python code. Feature extraction follows the Image Biomarker Standardization Initiative (IBSI) standards. MEDiml is built around complexity minimization: features are grouped into levels of increasing complexity (morphological, intensity, textural,l and filter-based), a model is trained at each level, and the simplest level not significantly outperformed by more complex ones is retained. Applied to five oncological datasets (2,104 scans), MEDiml achieved AUCs of 0.65 to 0.82, with morphological or intensity features sufficient in four of the five tasks. Resources: www.mediml.app.
Abstract Selective removal of functional groups offers a powerful route to streamline molecular design and generate structural diversity. Th… (see more)e use of such deletion strategies has surged in organic synthesis, yet achieving these transformations with high selectivity remains a formidable challenge. Here, we present a photoenzymatic platform for stereocontrolled C–CN bond cleavage of malononitriles, enabling access to diverse chiral nitriles. Directed evolution of the flavin-dependent enzyme OYE1 yielded a variant that efficiently converted a wide range of geminal dinitriles, affording products in high yields (up to >99%) and enantioselectivities (up to 3:97 e.r.). The platform was further extended to a one-pot cascade combining Michael addition with enzymatic decyanation. Mechanistic studies revealed a pathway involving a benzyl radical intermediate (Int1), followed by a carbanion species (Int2). This work establishes the first photobiocatalytic cyano group deletion strategy and provides a modular approach for sustainable synthesis of valuable chiral nitrile scaffolds.
2026-10-26
Journal of the American Chemical Society (published)
Microlensing of strongly lensed quasars probes unresolved accretion disks, but many inference methods use static magnification maps with fix… (see more)ed relative microlens positions. Using the GPU-accelerated simulator Microcaustics, we generate ten-year multiband light curves for three quadruply lensed quasar configurations. We compare dynamical microlensing with independent microlens motions against coherent bulk motion, and a microlens mass distribution against equal masses, keeping the velocity distribution of individual microlenses and mean mass fixed. We infer population disk sizes using ridge regression on the means and lower and upper scatters of ten variability and event metrics. We measure biases by applying models trained with simplified physics to simulations including both effects. Models trained on coherent motion interpret the additional variability from evolving caustics as smaller disks. For populations of 128 independent realizations, jointly inferred radii are consistently smaller by 6-28%. Microlens dynamics increase caustic-crossing rates pooled across all four macroimages by 5-15% and r-band high-magnification-event (HME) rates by 13-25%, while shortening median HME durations by 4-12%. Omitting the mass distribution produces configuration-dependent disk-size offsets from -12% to +5%. Including the mass distribution decreases caustic-crossing rates by 6-8% and HME rates by 6-12%. It also narrows the spread of ten-year averaged microlensing time-delay differences between macroimages by a median of 9-12% per system, while microlens dynamics have a smaller, image-pair-dependent effect. These results demonstrate the importance of microlens dynamics for population disk-size inference and motivate simulations that include dynamical microlensing and a microlens mass distribution.
Modern machine learning models are often trained on large datasets assembled from many different sources, whose statistical properties can d… (see more)iffer substantially. Understanding how this heterogeneity affects generalization performance is therefore important for explaining how model performance changes as model size increases. In this work, we study this question in linear random-feature models and show that mixtures of datasets with different covariance structures can give rise to a range of multiple-descent phenomena. We derive exact asymptotic expressions for the test error of each data source in the high-dimensional regime, and show how heterogeneity in covariance scale and spectral structure can introduce additional, group-dependent interpolation thresholds as model size increases. These phenomena arise even with noiseless labels and linear activations, showing that neither label noise nor activation nonlinearity is necessary for multiple-descent. Our results therefore provide a different perspective from classical explanations of multiple descent. More broadly, our results have potential implications for algorithmic fairness by showing that model scaling can affect different data groups unevenly, amplifying disparities in their generalization errors. We validate our theoretical results on synthetic Gaussian mixtures and real-world datasets, including MNIST and CIFAR-10.
Neural music editing requires a system to perform a requested change while preserving the musical content that should remain unchanged. Exis… (see more)ting evaluations often lack a paired target edit, making these requirements difficult to measure separately. We introduce SMEBench, a large-scale benchmark of controlled music edits constructed by modifying symbolic MIDI performances and rendering matched source and target audio. Each example contains a source, an instruction, a paired target, edit metadata, and complementary descriptions. The benchmark covers twelve instrumentation, articulation, dynamics, and pitch operations and includes a balanced 3,600-example test set together with a smaller real-audio complement derived from multitrack recordings. The paired targets allow source preservation, target agreement, and textual compatibility to be evaluated independently. We use these references to diagnose common audio-editing metrics and benchmark six representative editing systems, showing that the metric used to rank a system can materially change the conclusion. The dataset is available at our repository.
In drug discovery, quantitative structure–activity relationship (QSAR) models are widely used to guide Go/No-Go decisions within the Desig… (see more)n–Make–Test–Analyze (DMTA) cycle. However, conventional decision heuristics typically rely on a single cutoff, leading to a rigid binary select/discard paradigm. This approach is particularly ill-suited for borderline compounds near the decision boundary, where screening decisions are especially sensitive to prediction uncertainty and premature choices may either discard viable leads or advance likely failures, thereby increasing downstream assay costs. To address this limitation, we propose Regional Selection (RS), an uncertainty-aware three-way decision framework that partitions compounds into Predicted Pass, Predicted Fail, and Predicted Indeterminate regions. By explicitly reserving high-uncertainty compounds for targeted follow-up, RS avoids the pitfalls of premature binary classification. We formalize this framework through Regional Selection Inference (RSI), which casts region assignment as a multiple-hypothesis testing problem. We develop two imple- mentations of RSI: an empirical calibration-based method (RSI-EC), which thresholds uncertainty-normalized scores via empirical calibration, and a conformal selectionbased method (RSI-CS), which constructs conformal p-values for region assignment. RSI-EC is supported by large-sample calibration arguments, whereas RSI-CS provides finite-sample, distribution-free guarantees under exchangeability. Extensive evaluations across 15 high-dimensional QSAR benchmarks show that both RSI procedures reliably control the false discovery rate while maintaining high screening power. In limited-data regimes, RSI-CS yields particularly stable FDR control, whereas RSI-EC can be slightly less conservative; both perform strongly as sample sizes increase. We further study a cost-aware extension that incorporates asymmetric downstream costs through the score construction while keeping the nominal FDR target fixed. This extension introduces a tuning parameter that can reduce realized downstream cost, with dataset-dependent trade-offs against screening power. Overall, RSI offers a mathematically grounded and resource-aware alternative to single-threshold screening, allowing discovery teams to better balance decision confidence with assay budgets.
Abstract Foundation models have driven major advancements in the language and vision domains, as seen in the popularity of Artificial Intell… (see more)igence (AI) tools, and a growing number of studies are bringing the approach to human brain imaging. Whether these models are ready for independent researchers to use on clinical datasets has yet to be examined. We evaluated two recent fMRI foundation models, BrainLM and BrainHarmonix, on PreventAD, a dataset on adults with a family history of Alzheimer’s disease. We compared features from frozen and fine-tuned models against two baselines: functional connectivity and a dummy classifier that always predicts the most common class. The sample includes 343 subjects who passed quality control, with a subset of 218 who had amyloid PET imaging. We tested four prediction targets: sex, age, progression to mild cognitive impairment (MCI), and β-amyloid burden. Both models required substantial changes to the released code. BrainLM’s own data preparation workflow did not produce stable fine-tuning, which we achieved only after switching to a standard workflow implemented with the library Nilearn. BrainHarmonix’s documentation did not specify how to normalize new data, so we resolved this by testing different alternatives. For sex, functional connectivity outperformed all foundation model features. For age group, the best foundation model feature, taken from BrainHarmonix’s structural (T1-weighted) encoder, slightly exceeded functional connectivity, but the difference was not significant. For MCI and β-amyloid, no feature exceeded the dummy classifier in accuracy, and the area under the curve (AUC) stayed near chance for all features. For both clinical targets, the most precise feature came from fine-tuned BrainLM, meaning a larger share of its positive predictions were correct than for functional connectivity, but neither difference reached statistical significance. These results suggested that the strong transfer to new tasks seen in large language models cannot yet be assumed for fMRI foundation models, and that both models required code adaptation that would preclude their streamlined adoption by independent groups. We recommend that future studies provide standardized data workflows and always report functional connectivity and dummy classifier baselines. For clinical targets with strong class imbalance, studies should report several metrics, including precision.