Existing weakly supervised temporal action localization (WTAL) methods typically follow a decoupled classification-localization pipeline: segment-level classifiers are trained first, and their predictions are then aggregated to score proposals at inference. Under this training-inference discrepancy, proposal scoring at inference relies on an additional aggregation step, which can accumulate errors from noisy segment responses and thus undermine score reliability. Moreover, proposal scores are often directly used as confidence without explicit score-quality modeling or quality-aware evaluation, further contributing to pronounced score-quality misalignment and thus widening the classification-localization gap. To address proposal score-quality misalignment, we propose ACL-Net, a framework for proposal score calibration. At its core is a dual-axis Proposal-level Action Consistency Learning (PACL) paradigm, implemented through two complementary modules: (i) a Semantic Consistency Module (SCM) that refines proposal representations by maintaining fused class centers to enforce compact and robust same-class features; within SCM, a cross-modal consistency-driven Classification Enhancement Module (CEM) denoises the fused class centers to mitigate error accumulation under weak supervision; and (ii) a Process Consistency Module (PCM) that derives geometry-aware reference scores from relative temporal relations among overlapping proposals, guiding the model to assess proposal quality in terms of relative process completeness and improve score-quality alignment. By jointly modeling semantic and process consistency to calibrate proposal scores, ACL-Net markedly improves localization accuracy. On THUMOS14 and ActivityNet1.3, it achieves state-of-the-art performance with uniform and substantial gains across multiple established baselines, while markedly lowering the expected calibration error (ECE).

Next activity prediction, a critical subtask of Predictive Process Monitoring (PPM), aims to forecast the upcoming activity of an ongoing process instance based on event logs generated by process-aware information systems, thereby enabling businesses to proactively mitigate potential risks. While recent deep sequence models have shown strong predictive capabilities, they often fail to explicitly capture the intrinsic dependencies that govern the conditions and impacts of event occurrences in business processes. To address this limitation, we propose a novel Place-Transition Embedding (PTE) framework that captures these dependencies through Petri-net–inspired token consumption and generation semantics. PTE employs a deep embedding network to simulate the evolution of markings, i.e., how each observed event drives token dynamics over places over time under our resource interpretation. In addition, we introduce a temporal scoring mechanism that ranks candidate activities by jointly considering resource feasibility and temporal urgency, improving both interpretability and predictive accuracy. Extensive experiments on six real-world datasets show that PTE achieves state-of-the-art performance for next-activity prediction. Moreover, PTE generalizes robustly under sparse training conditions, highlighting its practical applicability in real-world scenarios with limited business data.

Weakly-supervised temporal action localization (WTAL) aims to identify and localize action instances in untrimmed videos using only video-level labels. Existing methods typically rely on original features from frozen pre-trained encoders designed for trimmed action classification (TAC) tasks, which inevitably introduces task discrepancy. Additionally, these methods often overlook the importance of considering action consistency from multiple perspectives, specifically the consistency in action processes and action semantics, both of which are crucial for the model's understanding of actions. To address these issues, we propose a novel WTAL method based on similar modality enhancement and action consistency learning (SEAL). First, we construct global descriptors for each action category, and use the pseudo-labels generated based on these descriptors to guide the model in learning more consistent representations, thereby mitigating task discrepancy. Second, we design two types of losses to achieve action consistency learning: process consistency loss, which penalizes candidate proposals that deviate from the action center to ensure the completeness of the action process, and semantic consistency loss, which employs local descriptors to help proposals of the same action category (especially those with apparent semantic confusion) learn similar feature distributions. Extensive experiments on the THUMOS14 and ActivityNet datasets demonstrate the superior performance of the proposed method compared to state-of-the-art methods.

Predictive process monitoring, a technique within the field of process mining, aims primarily to forecast the next execution activity of ongoing traces. This capability enables organizations to proactively discover anomalies and enhance intelligent decision-making in business processes. Existing predictive methods only consider the individual context of traces or the structural constraints of process models, failing to comprehensively describe the state of ongoing traces. Therefore, this paper proposes SGAP (Sequential Graph for next Activity Prediction), a predictive model that combines contextual information within traces and collaborative behaviors among traces. In SGAP, sequential graphs are constructed by transforming activities and traces from event logs into nodes and edges, which offers a flexible framework for understanding the impact of various trace contexts on predictions. Moreover, to adaptively handle the message passing across multiple and heterogeneous edges in sequential graphs, we devise a trace attention mechanism that considers different sequential patterns of edges. Experimental results conducted over six real-world datasets demonstrate the effectiveness of sequential graph modeling on trace contexts and highlight the significance of SGAP in forecasting the next activities.