Remi Lam, Alvaro Sanchez-Gonzalez, Matthew Willson
et al.
Global medium-range weather forecasting is critical to decision-making across many social and economic domains. Traditional numerical weather prediction uses increased compute resources to improve forecast accuracy but does not directly use historical weather data to improve the underlying model. Here, we introduce GraphCast, a machine learning–based method trained directly from reanalysis data. It predicts hundreds of weather variables for the next 10 days at 0.25° resolution globally in under 1 minute. GraphCast significantly outperforms the most accurate operational deterministic systems on 90% of 1380 verification targets, and its forecasts support better severe event prediction, including tropical cyclone tracking, atmospheric rivers, and extreme temperatures. GraphCast is a key advance in accurate and efficient weather forecasting and helps realize the promise of machine learning for modeling complex dynamical systems. Editor’s summary The numerical models used to predict weather are large, complex, and computationally demanding and do not learn from past weather patterns. Lam et al. introduced a machine learning–based method that has been trained directly from reanalysis data of past atmospheric conditions. In this way, the authors were able to quickly predict hundreds of weather variables globally up to 10 days in advance and at high resolution. Their predictions were more accurate than those of traditional weather models in 90% of tested cases and displayed better severe event prediction for tropical cyclones, atmospheric rivers, and extreme temperatures. —H. Jesse Smith Machine learning leads to better, faster, and cheaper weather forecasting.
Deep learning tools have gained tremendous attention in applied machine learning. However such tools for regression and classification do not capture model uncertainty. In comparison, Bayesian models offer a mathematically grounded framework to reason about model uncertainty, but usually come with a prohibitive computational cost. In this paper we develop a new theoretical framework casting dropout training in deep neural networks (NNs) as approximate Bayesian inference in deep Gaussian processes. A direct result of this theory gives us tools to model uncertainty with dropout NNs -- extracting information from existing models that has been thrown away so far. This mitigates the problem of representing uncertainty in deep learning without sacrificing either computational complexity or test accuracy. We perform an extensive study of the properties of dropout's uncertainty. Various network architectures and non-linearities are assessed on tasks of regression and classification, using MNIST as an example. We show a considerable improvement in predictive log-likelihood and RMSE compared to existing state-of-the-art methods, and finish by using dropout's uncertainty in deep reinforcement learning.
Jascha Narain Sohl-Dickstein, Eric A. Weiss, Niru Maheswaranathan
et al.
A central problem in machine learning involves modeling complex data-sets using highly flexible families of probability distributions in which learning, sampling, inference, and evaluation are still analytically or computationally tractable. Here, we develop an approach that simultaneously achieves both flexibility and tractability. The essential idea, inspired by non-equilibrium statistical physics, is to systematically and slowly destroy structure in a data distribution through an iterative forward diffusion process. We then learn a reverse diffusion process that restores structure in data, yielding a highly flexible and tractable generative model of the data. This approach allows us to rapidly learn, sample from, and evaluate probabilities in deep generative models with thousands of layers or time steps, as well as to compute conditional and posterior probabilities under the learned model. We additionally release an open source reference implementation of the algorithm.
The encoder-decoder framework for neural machine translation (NMT) has been shown effective in large data scenarios, but is much less effective for low-resource languages. We present a transfer learning method that significantly improves Bleu scores across a range of low-resource languages. Our key idea is to first train a high-resource language pair (the parent model), then transfer some of the learned parameters to the low-resource pair (the child model) to initialize and constrain training. Using our transfer learning method we improve baseline NMT models by an average of 5.6 Bleu on four low-resource language pairs. Ensembling and unknown word replacement add another 2 Bleu which brings the NMT performance on low-resource machine translation close to a strong syntax based machine translation (SBMT) system, exceeding its performance on one language pair. Additionally, using the transfer learning model for re-scoring, we can improve the SBMT system by an average of 1.3 Bleu, improving the state-of-the-art on low-resource machine translation.
While neural machine translation (NMT) is making good progress in the past two years, tens of millions of bilingual sentence pairs are needed for its training. However, human labeling is very costly. To tackle this training data bottleneck, we develop a dual-learning mechanism, which can enable an NMT system to automatically learn from unlabeled data through a dual-learning game. This mechanism is inspired by the following observation: any machine translation task has a dual task, e.g., English-to-French translation (primal) versus French-to-English translation (dual); the primal and dual tasks can form a closed loop, and generate informative feedback signals to train the translation models, even if without the involvement of a human labeler. In the dual-learning mechanism, we use one agent to represent the model for the primal task and the other agent to represent the model for the dual task, then ask them to teach each other through a reinforcement learning process. Based on the feedback signals generated during this process (e.g., the language-model likelihood of the output of a model, and the reconstruction error of the original sentence after the primal and dual translations), we can iteratively update the two models until convergence (e.g., using the policy gradient methods). We call the corresponding approach to neural machine translation dual-NMT. Experiments show that dual-NMT works very well on English ↔ French translation; especially, by learning from monolingual data (with 10% bilingual data for warm start), it achieves a comparable accuracy to NMT trained from the full bilingual data for the French-to-English translation task.
During the past decade, virtual screening (VS) has evolved from traditional similarity searching, which utilizes single reference compounds, into an advanced application domain for data mining and machine-learning approaches, which require large and representative training-set compounds to learn robust decision rules. The explosive growth in the amount of public domain-available chemical and biological data has generated huge effort to design, analyze, and apply novel learning methodologies. Here, I focus on machine-learning techniques within the context of ligand-based VS (LBVS). In addition, I analyze several relevant VS studies from recent publications, providing a detailed view of the current state-of-the-art in this field and highlighting not only the problematic issues, but also the successes and opportunities for further advances.
Recommender systems use algorithms to provide users with product or service recommendations. Recently, these systems have been using machine learning algorithms from the field of artificial intelligence. However, choosing a suitable machine learning algorithm for a recommender system is difficult because of the number of algorithms described in the literature. Researchers and practitioners developing recommender systems are left with little information about the current approaches in algorithm usage. Moreover, the development of a recommender system using a machine learning algorithm often has problems and open questions that must be evaluated, so software engineers know where to focus research efforts. This paper presents a systematic review of the literature that analyzes the use of machine learning algorithms in recommender systems and identifies research opportunities for software engineering research. The study concludes that Bayesian and decision tree algorithms are widely used in recommender systems because of their relative simplicity, and that requirement and design phases of recommender system development appear to offer opportunities for further research.
Justice Kwame Appati, Alfred Tettey Ternor, Waliyyullah Umar Bandawu
et al.
Abstract Parkinson’s Disease (PD) is a progressive neurological disorder affecting motor and non-motor functions. Early detection significantly improves patient outcomes, yet traditional clinical diagnoses are often delayed. Machine learning (ML), especially using speech data—given that over 90% of PD patients experience speech impairments—offers a promising alternative for early diagnosis. However, the high dimensionality of PD datasets poses challenges for prediction accuracy, highlighting the need for effective feature selection. This study proposes a novel hybrid feature selection method, the Chaotic Grey Wolf–Dragonfly Algorithm (CGWO-DA), which integrates the Grey Wolf Optimizer (GWO), Dragonfly Algorithm (DA), and a Logistic Chaotic Map to improve the balance between exploration and exploitation and prevent premature convergence. CGWO-DA was applied to three PD speech datasets of varying sizes. Preprocessing steps included label encoding, normalization, and irrelevant column removal, followed by an 80–20 training-test data split. CGWO-DA outperformed traditional methods, selecting optimal features and improving classifier performance. On a small dataset, it achieved 100% accuracy using Random Forest with 13 selected features. On medium and large datasets, it achieved 90% and 96% accuracy using Deep Neural Networks and Random Forest, respectively. These findings highlight CGWO-DA’s effectiveness and its potential for broader application, including the diagnosis of non-motor PD symptoms.
Machine learning approaches are increasingly successful in image-based diagnosis, disease prognosis, and risk assessment. This paper highlights new research directions and discusses three main challenges related to machine learning in medical imaging: coping with variation in imaging protocols, learning from weak labels, and interpretation and evaluation of results.