iTransformer
PulseAugur coverage of iTransformer — every cluster mentioning iTransformer across labs, papers, and developer communities, ranked by signal.
5 day(s) with sentiment data
-
New JEPA model tackles cross-machine industrial AI transfer
Researchers have developed a Schema-Adaptive Action-Conditioned JEPA (SAAC-JEPA) designed for transferring industrial world models between different CNC machines. This model addresses challenges like differing dynamics,…
-
WaveHiTS model enhances wind direction forecasting with wavelet and hierarchical methods
A new model called WaveHiTS has been proposed for wind direction forecasting, integrating wavelet transform with a hierarchical time series approach. This method decomposes wind direction into U-V components and uses wa…
-
New RATL method improves time-series forecasting using retrieved residuals
Researchers have developed RATL, a novel method for robust multivariate time-series forecasting that leverages retrieved historical forecast residuals. Unlike traditional approaches that discard residuals, RATL stores t…
-
New AdaRDiff method enhances time series forecasting accuracy
Researchers have developed AdaRDiff, a novel adaptive differencing method designed to improve long-horizon time series forecasting. This approach uses learnable weights to simplify series by subtracting weighted past va…
-
Neuro-symbolic framework predicts academic risk with interpretable AI
Researchers have developed EduRiskX, a novel neuro-symbolic framework designed to predict academic risk in online education. This system combines a Transformer-based neural network for analyzing student activity sequenc…
-
New multimodal AI framework improves influenza forecasting accuracy
Researchers have developed a novel multimodal deep learning framework called Dual-Stream Attention (DSA) for forecasting influenza-like illness (ILI) up to 12 weeks in advance. This framework effectively integrates nume…
-
Edge-AI system MotoSafety assesses two-wheeler collision risk
Researchers have developed MotoSafety, a novel edge-AI architecture designed to assess collision risk for two-wheeler riders under time pressure. This system utilizes a large dataset of multivariate time-series sequence…
-
AI model output heads more critical than backbones for financial forecasting
A new research paper suggests that for deep forecasting pipelines dealing with fat-tailed financial returns at short horizons, the output head of the model is more critical than the backbone architecture. Experiments co…
-
New Temporal Operator Attention framework enhances time-series analysis
Researchers have introduced Temporal Operator Attention (TOA), a novel framework designed to improve time-series analysis by addressing limitations in standard attention mechanisms. TOA explicitly incorporates learnable…
-
Machine learning models show promise for Bitcoin trading after costs
A new research paper explores the use of machine learning models, including XGBoost, LSTM, and iTransformer, for predicting Bitcoin returns. The study found that while these models can generate positive gross trading pe…
-
HEPA architecture predicts critical time-series events using self-supervision
Researchers have developed HEPA, a novel self-supervised architecture for predicting critical events in multivariate time series data. This architecture uses a causal Transformer encoder pretrained with a Joint-Embeddin…
-
Deep learning framework predicts adaptive alarm thresholds for 4G networks
Researchers have developed a deep learning framework to automatically predict alarm thresholds for 4G mobile networks, aiming to improve service quality and reduce unnecessary engineer callouts. The proposed PCTN model …
-
New AI methods enhance time series forecasting accuracy and interpretability
Researchers have introduced several new methods for time-series forecasting, aiming to improve accuracy and generalization. MeLISA, a latent-free autoregressive model, enhances rollout efficiency and long-horizon statis…
-
DecompKAN model offers transparent, accurate long-term time series forecasting
Researchers have introduced DecompKAN, a novel architecture for long-term time series forecasting that prioritizes both predictive accuracy and model interpretability. This lightweight, attention-free system integrates …
-
Researchers use Transformers to generate reactive human motion from interaction data
Researchers have developed Transformer-based models to generate human motion in interactive scenarios, focusing on how one person's movement influences another's. They created a dataset from boxing videos to train and c…