To understand why this matters, it helps to start with the quantity being measured. The ionosphere forms where solar and cosmic radiation strip electrons from atoms in the mesosphere, thermosphere and exosphere, creating a plasma of free electrons and ions. The Total Electron Content, or TEC, counts the total number of electrons along a signal path, typically between a Global Navigation Satellite System satellite and a ground receiver. Because GPS and other positioning signals must traverse this charged region, TEC is both a nuisance for navigation accuracy and a remarkably sensitive diagnostic of what is happening overhead. Variations in TEC are driven primarily by space weather, including solar wind, solar flux and geomagnetic storms, but a growing body of research on Lithosphere-Atmosphere-Ionosphere Coupling, or LAIC, suggests that large earthquakes may perturb the ionosphere as well, through mechanisms that propagate energy and electric charge upward from the fault zone before and during rupture.

The idea that earthquakes leave traces in the sky is not new, but it remains contentious. Preparation zones around large faults can span hundreds of kilometers, and proposed coupling mechanisms range from acoustic-gravity waves launched by ground motion to electromagnetic effects associated with stress accumulation and radon release. What has been missing is a rigorous way to separate the seismic component of ionospheric variability from the much larger space-weather background. This is precisely the gap the new study addresses. Rather than simply hunting for anomalies, the researchers built predictive models of TEC using only solar and geomagnetic inputs, then evaluated how well those models captured the TEC behavior observed around the three Japanese earthquakes. Any systematic shortfall or distinctive pattern in the residuals becomes evidence of ionospheric disturbance tied to the seismic events themselves.

The data backbone of the study comes from three Japanese GNSS stations: USUD, AIRA and ISHI. True TEC values for these stations were obtained from IONOLAB, a well-established service for automatic near-real-time estimation of GPS-derived TEC. Solar and geomagnetic drivers were drawn from NASA’s OMNIWeb database and included the solar wind speed, the F10.7 solar flux index, the Disturbance Storm Time index known as Dst, and the Ap index of geomagnetic activity. Earthquake details were compiled from the United States Geological Survey. Together these inputs form a multivariate prediction problem: given the current state of the Sun and the geomagnetic field, what should the TEC be at each station? Deviations between prediction and observation during the earthquake windows then become the object of analysis.

Three machine learning architectures formed the core of the comparison. The first was a Gated Recurrent Unit network, a recurrent neural network designed to retain temporal dependencies in sequential data, trained with the Root Mean Square Propagation optimizer, or RMSProp, which adapts learning rates per parameter and is well suited to the noisy gradients typical of geophysical time series. The second was a Random Forest, an ensemble of decision trees introduced by Leo Breiman in 2001, whose hyperparameters were optimized using a Particle Swarm Optimizer, a bio-inspired search method that mimics the social behavior of bird flocks to explore the parameter space efficiently. The third was XGBoost, the extreme gradient boosting algorithm of Chen and Guestrin, optimized here with a Genetic Algorithm, an evolutionary search technique that iteratively selects, crosses and mutates candidate hyperparameter configurations. These three were benchmarked against NeQuick, a physics-based empirical model of the ionosphere widely used in the GNSS community.

Evaluation rested on four complementary statistical metrics: Root Mean Square Error, Mean Absolute Error, Mean Absolute Percentage Error and the Symmetric Mean Absolute Percentage Error. RMSE penalizes large errors heavily, MAE treats all errors equally, and the percentage-based measures express accuracy relative to the magnitude of the true values, which matters when TEC varies strongly with latitude, season and solar activity. The authors also performed cross-validation on Global Ionospheric Map data for the two best-performing machine learning models, a step that guards against the possibility that strong results reflect overfitting to a single dataset rather than genuine predictive skill. Cross-validation confirmed what the event-based comparisons suggested: XGBoost held its advantage when tested on data it had not been tuned against.

The headline result is that across all three earthquakes, the genetically optimized XGBoost model achieved the lowest RMSE and MAE values of any approach tested, consistently outperforming the GRU network, the swarm-optimized Random Forest and the NeQuick model. Among the remaining methods, Random Forest beat the GRU, an interesting outcome given that the recurrent network is explicitly designed for time series while the tree ensemble is not. The authors suggest that all models were able to capture TEC variations associated with the seismic events, which implies that the ionospheric response to these earthquakes is structured enough to be learned from solar and geomagnetic context, and that deviations from that learned baseline carry information about the lithosphere’s influence on the upper atmosphere.

There are several reasons why a gradient-boosted tree model might outperform a recurrent network in this setting. XGBoost builds an additive ensemble of shallow trees, each correcting the errors of its predecessors, and this architecture handles heterogeneous input features, such as the mix of solar wind speed, flux measurements and geomagnetic indices used here, without requiring the inputs to be sequenced or normalized to the same temporal scale. Boosting also tends to be robust with modest training sets, which is a real constraint in earthquake studies where the number of usable large events is inherently limited. The genetic algorithm’s global search over hyperparameters, including tree depth, learning rate and regularization strength, may have found configurations that gradient-based tuning would miss. The GRU, by contrast, must learn its temporal filters from data, and with only a handful of earthquake windows available for training, the recurrent model may simply have had too little sequence data to exploit its architectural advantage.

The broader significance of the work lies in the convergence of two research communities. Space-weather modelers have long sought better TEC prediction, because unmodeled ionospheric delay is one of the largest error sources in precise positioning and satellite navigation. Seismologists, meanwhile, have spent decades searching for reliable precursory signals, and ionospheric anomalies have been reported before earthquakes in Japan, Peru, Türkiye, Morocco, Cyprus, Alaska and the Himalayas, using ground GNSS networks and satellites such as DEMETER. By showing that machine learning models trained on space-weather inputs can characterize the expected ionospheric state with high fidelity, the study provides a cleaner statistical framework for asking whether specific anomalies exceed what space weather alone can explain. The LAIC hypothesis gains a sharper test, and navigation science gains better models in the same stroke.

Cautions remain, and the authors are careful not to overclaim. Three earthquakes, however well studied, do not establish a universal precursor signature, and the field has a long history of anomalies that looked compelling for one event but failed to generalize. The coupling physics connecting fault rupture to ionospheric electron content is still debated, and confounding factors such as volcanic activity, which Japan has in abundance, complicate attribution. What this study demonstrates is methodological progress: a validated, cross-checked machine learning pipeline that predicts TEC more accurately than a standard empirical model during seismic episodes, with XGBoost and genetic optimization at the top of the leaderboard. If future work extends this framework to more events, more regions and longer archives of GNSS data, the dream of reading warning signs from the ionosphere before the ground moves may edge closer to operational reality. For now, the sky above Japan has proven to be a measurable, modelable witness to the violence below.

Subject of Research: Machine learning analysis and prediction of ionospheric TEC anomalies associated with the 2024 Japan earthquakes

Article Title: Analysis of ionospheric TEC anomalies and prediction using XGBoost, GRU and RF algorithms associated with 2024 Japan Earthquakes (Mw > 7.0)

Article References: Mukesh, R., Kiruthiga, S., Rubashri, J., Fathima, S. R., Priyavarshini, P., Dass, S. C., Sahanaa, A. R. S., Safana, S., Karthick, S., & Ratnam, D. V. (2026). Analysis of ionospheric TEC anomalies and prediction using XGBoost, GRU and RF algorithms associated with 2024 Japan Earthquakes (Mw > 7.0). Earth Science Informatics, 19(11), Article 184. https://doi.org/10.1007/s12145-026-02220-9

Image Credits: AI Generated

DOI: 10.1007/s12145-026-02220-9

Keywords: ionosphere, total electron content, earthquake precursors, XGBoost, genetic algorithm, GRU, Random Forest, PSO, RMSProp, NeQuick, Japan earthquakes, lithosphere-atmosphere-ionosphere coupling

The research, published in the Journal of Ambient Intelligence and Humanized Computing, was conducted by Hossein Abbasimehr of Azarbaijan Shahid Madani University, Reza Paki of Politecnico di Milano, and Hamidreza Asadian Rad of Iran University of Science and Technology. Their framework, called Continual Forecasting Fusion Deep Reinforcement Learning, or CFFDRL, embeds streaming continual learning directly into the pipeline of a trading agent. The central idea is deceptively simple: instead of treating market prediction and trading decision-making as two frozen stages, the framework lets the forecasting component adapt continuously to newly generated data, so that the reinforcement learning agent always acts on a view of the market that reflects its most recent behavior.

Deep reinforcement learning has become one of the most actively explored approaches in algorithmic trading. In a typical setup, an agent observes the state of the market, takes actions such as buying, selling, or holding, and receives rewards tied to profit or risk-adjusted performance. Over many training episodes, the agent learns a policy that maps market states to actions. The problem, the authors note, is that these systems are usually optimized on historical data and then deployed as static models. When the underlying data-generating process changes, a phenomenon known in machine learning as concept drift, the learned policy can degrade badly. A policy tuned to a bull market may hold losing positions through a regime change; a strategy tuned to low volatility may misjudge risk when turbulence returns.

To combat this, the researchers turned to streaming continual learning, a branch of machine learning concerned with models that learn from an unbounded flow of data without forgetting what they already know. The specific technique at the heart of CFFDRL is Continuous Piggyback, an approach that adapts to newly generated data by learning task-specific masks over a frozen pre-trained backbone network, without modifying the original weights. Rather than retraining an entire neural network each time new data arrives, which is computationally expensive and risks erasing previously learned knowledge, the framework learns lightweight binary masks that select and reconfigure pathways through the frozen network for each new forecasting task. The result is a model that can absorb new market conditions while preserving the general structure it learned earlier.

The authors implemented this concept inside a gated recurrent unit, a type of recurrent neural network well suited to sequential data such as prices. The resulting module, called cPB-GRU, incrementally predicts future prices from historical OHLC data, the open, high, low, and close values that form the basic vocabulary of market analysis. Crucially, the module is continuously updated during both training and testing. This means the forecasting component does not stop learning when the evaluation phase begins; it keeps adapting as fresh market observations stream in, mirroring the way a human trader might recalibrate expectations day after day.

The forecasts generated by the cPB-GRU module are then concatenated with the raw OHLC data to form the observation space of the reinforcement learning agent. In other words, the trading agent does not only see what has happened in the market; it also sees a continuously refreshed estimate of what the forecasting module expects to happen next. This fusion of prediction and decision-making is what gives CFFDRL its name and its edge. The agent uses the proximal policy optimization algorithm, a widely used and stable reinforcement learning method, and benefits from observations that stay informative even as the market shifts beneath it.

The experimental evidence is drawn from six datasets, giving the comparison a breadth that single-asset backtests often lack. Across those datasets, CFFDRL achieved an average cumulative return of 50.09 percent, compared with 33.28 percent for a standard DRL-PPO baseline and 19.48 percent for a PPO variant paired with a static GRU forecaster. The gap is striking: the continual forecasting agent delivered roughly one and a half times the average return of the standard PPO setup and more than two and a half times that of the static forecasting configuration. The comparison with PPO-Static-GRU is particularly telling, because it isolates the contribution of continual adaptation; the only substantive difference is whether the forecasting module keeps learning from new data.

Profit alone is not the whole story in trading research, and the framework also performed well on standard risk metrics. CFFDRL achieved the highest average Sharpe ratio among the evaluated PPO variants, at 0.10, indicating better risk-adjusted returns, and the lowest average maximum drawdown, at 19.46 percent. Maximum drawdown measures the largest peak-to-trough decline an account experiences, and a lower value signals that the strategy avoids the deepest losses, a property investors typically prize as much as raw profitability. Taken together, the results indicate that continual forecasting improves not only how much the agent earns but how smoothly and safely it earns it.

The broader significance of the work lies in its marriage of two research traditions that have largely developed in parallel. Continual learning researchers have built sophisticated techniques for adapting models to data streams while preventing catastrophic forgetting, but most of that work has focused on classification tasks. Reinforcement learning researchers, meanwhile, have built increasingly powerful trading agents, but often without addressing the non-stationarity of financial data head-on. By making the forecasting module a living, evolving component of the observation space, CFFDRL offers a template for how streaming continual learning can be folded into decision-making systems that operate in environments where yesterday’s patterns are never quite today’s.

There are, of course, limits to what any backtest can promise. Live trading introduces transaction costs, slippage, liquidity constraints, and execution delays that no simulation fully captures, and the authors’ study reports no datasets generated or analyzed beyond the reported experiments. Still, the message of the research is clear and likely to resonate across quantitative finance: in non-stationary environments, the ability to keep learning is not a luxury but a determinant of performance. As automated trading systems take on a growing share of global market activity, frameworks like CFFDRL point toward a generation of agents that treat change not as a threat to be endured but as information to be absorbed, one streamed data point at a time.

Subject of Research: A deep reinforcement learning trading framework using streaming continual learning to adapt forecasts to evolving financial markets

Article Title: A novel deep reinforcement learning framework with task-incremental continual forecasting for trading systems

Article References: Abbasimehr, H., Paki, R., & Asadian Rad, H. (2026). A novel deep reinforcement learning framework with task-incremental continual forecasting for trading systems. Journal of Ambient Intelligence and Humanized Computing. https://doi.org/10.1007/s12652-026-05132-0

Image Credits: AI Generated

DOI: 10.1007/s12652-026-05132-0

Keywords: deep reinforcement learning, algorithmic trading, continual learning, streaming learning, concept drift, financial forecasting, GRU, proximal policy optimization, Sharpe ratio, maximum drawdown, trading systems, machine learning

Researchers led by Agung Bella Putra Utama, Aji Prasetya Wibawa, and Anik Nur Handayani of Universitas Negeri Malang, together with Andrew Nafalski of Adelaide University, have introduced a deterministic duel-based imputation framework that puts forecasting performance itself in charge of deciding how missing values should be filled. The central insight is deceptively simple: most imputation methods are judged by how statistically similar their reconstructions are to the original data, not by whether they actually help a forecasting model make better predictions. A filled-in series can look numerically plausible on paper and still wreck a downstream forecast by smoothing away the very peaks and rhythms the model needs to learn.

The framework works like a structured tournament. For every missing observation, the system generates a set of candidate values using familiar baseline techniques, including mean, median, and mode substitution, K-Nearest Neighbors, Multiple Imputation by Chained Equations, and Last-Observation-Carried-Forward. Each candidate is then scored with a composite loss function that combines three complementary forecasting metrics: Mean Absolute Percentage Error, which captures proportional error; Root Mean Square Error, which penalizes large absolute deviations; and the coefficient of determination, or R-squared, which measures how much of the variance in the real series the reconstruction explains. The weights are set at 0.4, 0.4, and 0.2 respectively, so error-based criteria dominate while explanatory power still plays a supporting role.

What separates this approach from conventional candidate ranking is the way winners are chosen. Instead of collapsing every candidate into a single aggregated score and picking the top one, the framework stages deterministic pairwise duels. Candidates face each other one-on-one, and a sensitivity threshold of 0.001 prevents negligible score differences from flipping outcomes. After all comparisons, candidates are ranked by cumulative wins, and the top three advance to a final aggregation stage. Four aggregation modes are available: averaging the survivors for low-variance stability, taking the median for outlier robustness, taking the maximum to emphasize peak responsiveness, or a winner-take-all selection of the candidate with the lowest forecasting loss. Because every step follows fixed rules, the same input always produces the same output, a property the authors argue is essential for auditing, validation, and accountability in operational systems.

The team evaluated the framework on the Beijing PM2.5 dataset, a widely used benchmark of hourly meteorological and air-pollution measurements collected between 2010 and 2014. After cleaning and temporal alignment, 41,776 valid records remained out of 43,800, with roughly 2,068 missing values concentrated in the PM2.5 variable itself. Crucially, the missingness is not random: gaps cluster in cold seasons and during periods of rapid atmospheric change, exactly the conditions where forecasting matters most and where naive imputation does the most damage. The dataset includes dew point, temperature, pressure, cumulative wind speed, snowfall, rainfall, and temporal indicators as explanatory variables, with PM2.5 concentration serving as the prediction target.

The imputed datasets were then fed into four recurrent forecasting architectures trained under identical conditions: a vanilla Recurrent Neural Network, a Long Short-Term Memory network, a Bidirectional LSTM, and a Gated Recurrent Unit. Hyperparameters for all models were tuned with Particle Swarm Optimization, though the researchers are careful to note that this stochastic tuning happens entirely outside the imputation stage and does not compromise its determinism. The optimized configurations converged on surprisingly lightweight designs: three hidden layers of 24 neurons each, sigmoid activations, the Adam optimizer with mean squared error loss, a batch size of 32, 46 training epochs, and a dropout rate of 0.2.

The results are striking. Compared with conventional imputation methods, the duel-based framework reduced MAPE by up to 70 percent and RMSE by 12 to 15 percent, while maintaining R-squared values above 0.95 across all four architectures. The Mean aggregation variant delivered the strongest overall performance, preserving both proportional variation and the amplitude structure of the signal. Dropping missing observations entirely, by contrast, produced the worst forecasts, confirming that simply discarding incomplete rows destroys the temporal coherence recurrent models rely on. Even BRITS, a sophisticated deep-learning imputation method, showed less consistent gains, suggesting that reconstruction quality alone does not guarantee forecasting quality.

Statistical testing reinforced the case. Paired t-tests and Wilcoxon signed-rank tests comparing the proposed Mean variant against KNN, the strongest conventional baseline, produced p-values below 0.01 across all architectures, with each experiment repeated ten times under fixed random seeds. Visual analysis of predicted versus observed PM2.5 trajectories showed close alignment across smooth and volatile segments alike, with the Bi-LSTM achieving the highest R-squared values and the GRU delivering the most efficient runtime while matching LSTM accuracy. The framework’s computational overhead scales quadratically with the number of candidates but linearly with data size, and because the candidate set is small and fixed, runtimes remain predictable even as data volume grows.

To test generalizability, the researchers extended the evaluation beyond air quality to three additional domains: the Heart Disease dataset, which has weak temporal structure; the KEDS e-journal dataset, which shows irregular behavioral patterns and moderate sparsity; and the Sunspot dataset, which exhibits strong periodic behavior. The framework achieved lower MAPE and RMSE than baselines including drop-missing, mean imputation, KNN, and BRITS across all of them, with the largest gains appearing in datasets with strong sequential patterns such as Sunspot. The authors acknowledge limitations, including the quadratic cost of pairwise comparison for large candidate pools and the fixed forecasting horizons tested, and they point to cluster-based candidate reduction and online learning as future directions. But the broader message stands: imputation should not be a passive preprocessing chore. Treated as a forecasting-driven decision process, it becomes a strategic lever for building AI systems that are not only accurate but consistent, transparent, and worthy of trust.

Subject of Research: A deterministic duel-based imputation framework that integrates forecasting performance metrics into missing-data selection for reliable AI-driven time-series forecasting.

Article Title: A deterministic duel-based imputation framework for reliable AI-driven time-series forecasting

Article References: Utama, A. B. P., Wibawa, A. P., Handayani, A. N., & Nafalski, A. (2026). A deterministic duel-based imputation framework for reliable AI-driven time-series forecasting. Discover Informatics, 1(1), Article 3. https://doi.org/10.1007/s44564-026-00001-6

Image Credits: AI Generated

DOI: 10.1007/s44564-026-00001-6

Keywords: time-series imputation, missing data, AI forecasting, deterministic imputation, deep learning, LSTM, GRU, RNN, PM2.5, forecasting reliability, machine learning, data preprocessing