Long-term Dependencies Research Articles

Developing a Dynamic Understanding of Risk Factors for People Admitted to Long-term High Dependency Units

AbstractEven with developments in supporting people with enduring psychosis, some people living with these difficulties still require specialised care in inpatient rehabilitation facilities. To optimise the support provided a service evaluation was conducted for people admitted to a Long-term High Dependency Unit service. Data collected routinely with the Short-Term Assessment for Risk and Treatability tool was used to identify dynamic risk factors for the residents. Completed assessments were analysed for frequency and severity of risk behaviours; historic prevalence of risk behaviours; judgements on future risk; and strengths and vulnerabilities. The evaluation indicated a picture of a group of people who have a variety of historic risks and currently engage regularly in aggression and self-neglect, whilst experiencing limited insight, social exclusion, and limited coping abilities. Strengths and vulnerabilities linked to risk behaviours were also grouped into conceptually similar domains to aid intervention. Recommendations are made for using such data to enhance recovery.

Minutely multi-step irradiance forecasting based on all-sky images using LSTM-InformerStack hybrid model with dual feature enhancement

In recent years, the government has attached great importance to promoting innovation in new energy technologies, particularly in the area of solar power generation. Photovoltaic (PV) power generation has become an essential factor in the promotion of renewable energy due to its inherent safety, environmental sustainability and abundant solar energy resources. The efficiency and stability of PV generation systems are closely related to the accurate prediction of real-time global horizontal irradiance (GHI). However, some existing research methods have made some progress in hourly resolution, but the multi-step prediction accuracy and model generalization ability in high time resolution are still unsatisfactory. Therefore, this paper proposes a novel minutely multi-step irradiance forecasting based on all-sky images using LSTM-InformerStack hybrid model with dual feature enhancement scheme. In this scheme, the advantages of both LSTM and InformerStack models are combined to further enhance the global and local features, which can better capture the short-term time series patterns and long-term dependencies and improve the accuracy of the time series. The experimental validation results show that the performance of this prediction scheme is improved by 11.2%, and 26.0% for nRMSE and R2, respectively, and the 24-step prediction can be improved up to 26.1% vs. the persistent model. Meanwhile, the prediction scheme of this paper is also expected to provide a reference in future multi-step time series accurate prediction with higher resolution.

MOTC: Abdominal Multi-objective Segmentation Model with Parallel Fusion of Global and Local Information.

Convolutional Neural Networks have been widely applied in medical image segmentation. However, the existence of local inductive bias in convolutional operations restricts the modeling of long-term dependencies. The introduction of Transformer enables the modeling of long-term dependencies and partially eliminates the local inductive bias in convolutional operations, thereby improving the accuracy of tasks such as segmentation and classification. Researchers have proposed various hybrid structures combining Transformer and Convolutional Neural Networks. One strategy is to stack Transformer blocks and convolutional blocks to concentrate on eliminating the accumulated local bias of convolutional operations. Another strategy is to nest convolutional blocks and Transformer blocks to eliminate bias within each nested block. However, due to the granularity of bias elimination operations, these two strategies cannot fully exploit the potential of Transformer. In this paper, a parallel hybrid model is proposed for segmentation, which includes a Transformer branch and a Convolutional Neural Network branch in encoder. After parallel feature extraction, inter-layer information fusion and exchange of complementary information are performed between the two branches, simultaneously extracting local and global features while eliminating the local bias generated by convolutional operations within the current layer. A pure convolutional operation is used in decoder to obtain final segmentation results. To validate the impact of the granularity of bias elimination operations on the effectiveness of local bias elimination, the experiments in this paper were conducted on Flare21 dataset and Amos22 dataset. The average Dice coefficient reached 92.65% on Flare21 dataset, and 91.61% on Amos22 dataset, surpassing comparative methods. The experimental results demonstrate that smaller granularity of bias elimination operations leads to better performance.

Yield curve momentum

Abstract I analyze time series momentum along the Treasury term structure. Yield curve momentum is primarily due to changes in the level factor of yields. Because yield changes are partly induced by changes in the federal funds rate, yield curve momentum is related to post-FOMC (Federal Open Market Committee) announcement drift. The momentum factor is unspanned by the information in the term structure today and is hence inconsistent with standard term structure, macrofinance, and behavioral models. I argue that the results are consistent with a model with unpriced longer term dependencies.

Deep Learning-Based Managing Disaster with Smart Education in Environmental Sustainability while Avoiding Fake News

Networking continues to increase day by day to invest increasingly our daily, creating a huge volume of various and precise data: it is not easy to collect content, especially in crisis times. We focus on Smart Education proposed as the primary tool of a hybrid of Deep Convolutional Neural Networks (CNN)-Long Short-Term Memory (LSTM)-based model to retrieve content efficiently: CNN is used to extract meaningful features from multiple sources, enabling to have qualitative and sure information, notably with an efficient fake news security, and LSTM is used to maintain long-term dependencies in the extracted features with recurrent connections. This model has been compared to previous approaches to the performance of a publicly available dataset to demonstrate its highly satisfactory performance. This new approach makes it possible to integrate artificial intelligence technologies, deep learning, social media, and detecting or avoiding fake news into the crisis management model. It is based on an extension of our previous approach, namely disaster management based on short-term memory and education: this experience constitutes a background for this model. It combines representation training with awareness and education while retrieving pattern information by combining various search results from multiple sources. We have extended it to improve our disaster management model and evaluate it in the case of COVID-19 while obtaining promising results, through past programs and experiences that have shown overwhelmingly positive effects of education for vulnerability reduction and disaster risk management, in the pursuit of Environmental Sustainability

Hierarchical multi-head attention LSTM for polyphonic symbolic melody generation

Creating symbolic melodies with machine learning is challenging because it requires an understanding of musical structure and the handling of inter-dependencies and long-term dependencies. Learning the relationship between events that occur far apart in time in music poses a considerable challenge for machine learning models. Another notable feature of music is that notes must account for several inter-dependencies, including melodic, harmonic, and rhythmic aspects. Baseline methods, such as RNNs, LSTMs, and GRUs, often struggle to capture these dependencies, resulting in the generation of musically incoherent or repetitive melodies. As such, in this study, a hierarchical multi-head attention LSTM model is proposed for creating polyphonic symbolic melodies. This enables our model to generate more complex and expressive melodies than previous methods, while still being musically coherent. The model allows learning of long-term dependencies at different levels of abstraction, while retaining the ability to form inter-dependencies. The study has been conducted on two major symbolic music datasets, MAESTRO and Classical-Music MIDI, which feature musical content encoded on MIDI. The artistic nature of music poses a challenge to evaluating the generated content and qualitative analysis are often not enough. Thus, human listening tests are conducted to strengthen the evaluation. Qualitative analysis conducted on the generated melodies shows significantly improved loss scores on MSE over baseline methods, and is able to generate melodies that were both musically coherent and expressive. The listening tests conducted using Likert-scale support the qualitative results and provide better statistical scores over baseline methods.

Research on Financial Risk Intelligent Monitoring and Early Warning Model Based on LSTM, Transformer, and Deep Learning

As global financial markets continue to evolve and change, financial risk monitoring and early warning have become increasingly important. However, the complexity and diversity of financial markets have led to the emergence of multidimensional and multimodal data. Traditional risk monitoring methods face difficulties in handling such diverse data and adapting to the monitoring and early warning needs of emerging risk types. To address these issues, this article proposes a financial risk intelligent monitoring and early warning model that integrates deep learning to better cope with uncertainty and risk in the financial market. Firstly, the authors introduce an LSTM model in the initial approach, trained on historical financial market data, to capture long-term dependencies and trends in the data, enabling effective monitoring of financial risk. They also optimize the model architecture to improve its performance and prediction accuracy. Secondly, the authors further introduce a transformer model with self-attention mechanism to better handle sequential data.

Gated Fusion Adaptive Graph Neural Network for Urban Road Traffic Flow Prediction

Accurate prediction of traffic flow plays an important role in maintaining traffic order and traffic safety, which is a key task in the application of intelligent transportation systems (ITS). However, the urban road network has complex dynamic spatial correlation and nonlinear temporal correlation, and achieving accurate traffic flow prediction is a highly challenging task. Traditional methods use sensors deployed on roads to construct the spatial structure of the road network and capture spatial information by graph convolution. However, they ignore that the spatial correlation between nodes is dynamically changing, and using a fixed adjacency matrix cannot reflect the real road spatial structure. To overcome these limitations, this paper proposes a new spatial-temporal deep learning model: gated fusion adaptive graph neural network (GFAGNN). GFAGNN first extracts long-term dependencies on raw data through stacking expansion causal convolution, Then the spatial features of the dynamics are learned by adaptive graph attention network and adaptive graph convolutional network respectively, Finally the fused information is passed through a lightweight channel attention to extract temporal features. The experimental results on two public data sets show that our model can effectively capture the spatiotemporal correlation in traffic flow prediction. Compared with GWNET-conv model on METR-LA dataset, the three indexes in the 60-minute task prediction improved by 2.27%,2.06% and 2.13%, respectively.

Conversational recommender based on graph sparsification and multi-hop attention

Conversational recommender systems provide users with item recommendations via interactive dialogues. Existing methods using graph neural networks have been proven to be an adequate representation of the learning framework for knowledge graphs. However, the knowledge graph involved in the dialogue context is vast and noisy, especially the noise graph nodes, which restrict the primary node’s aggregation to neighbor nodes. In addition, although the recurrent neural network can encode the local structure of word sequences in a dialogue context, it may still be challenging to remember long-term dependencies. To tackle these problems, we propose a sparse multi-hop conversational recommender model named SMCR, which accurately identifies important edges through matching items, thus reducing the computational complexity of sparse graphs. Specifically, we design a multi-hop attention network to encode dialogue context, which can quickly encode the long dialogue sequences to capture the long-term dependencies. Furthermore, we utilize a variational auto-encoder to learn topic information for capturing syntactic dependencies. Extensive experiments on the travel dialogue dataset show significant improvements in our proposed model over the state-of-the-art methods in evaluating recommendation and dialogue generation.

Sub-convolutional U-Net with transformer attention network for end-to-end single-channel speech enhancement

Recent advancements in deep learning-based speech enhancement models have extensively used attention mechanisms to achieve state-of-the-art methods by demonstrating their effectiveness. This paper proposes a transformer attention network based sub-convolutional U-Net (TANSCUNet) for speech enhancement. Instead of adopting conventional RNNs and temporal convolutional networks for sequence modeling, we employ a novel transformer-based attention network between the sub-convolutional U-Net encoder and decoder for better feature learning. More specifically, it is composed of several adaptive time―frequency attention modules and an adaptive hierarchical attention module, aiming to capture long-term time-frequency dependencies and further aggregate hierarchical contextual information. Additionally, a sub-convolutional encoder-decoder model used different kernel sizes to extract multi-scale local and contextual features from the noisy speech. The experimental results show that the proposed model outperforms several state-of-the-art methods.

Local-Global Spatial-Temporal Graph Convolutional Network for Traffic Flow Forecasting

Traffic forecasting’s key challenge is to extract dynamic spatial-temporal features within intricate traffic systems. This paper introduces a novel framework for traffic prediction, named Local-Global Spatial-Temporal Graph Convolutional Network (LGSTGCN). The framework consists of three core components. Firstly, a graph attention residual network layer is proposed to capture global spatial dependencies by evaluating traffic mode correlations between different nodes. The context information added in the residual connection can improve the generalization ability of the model. Secondly, a T-GCN module, combining a Graph Convolution Network (GCN) with a Gated Recurrent Unit (GRU), is introduced to capture real-time local spatial-temporal dependencies. Finally, a transformer layer is designed to extract long-term temporal dependence and to identify the sequence characteristics of traffic data through positional encoding. Experiments conducted on four real traffic datasets validate the forecasting performance of the LGSTGCN model. The results demonstrate that LGSTGCN can achieve good performance and be applicable to traffic forecasting tasks.

A Deep Learning Framework for Predicting Patient Decannulation on Extracorporeal Membrane Oxygenation Devices: Development and Model Analysis Study.

Venovenous extracorporeal membrane oxygenation (VV-ECMO) is a therapy for patients with refractory respiratory failure. The decision to decannulate someone from extracorporeal membrane oxygenation (ECMO) often involves weaning trials and clinical intuition. To date, there are limited prognostication metrics to guide clinical decision-making to determine which patients will be successfully weaned and decannulated. This study aims to assist clinicians with the decision to decannulate a patient from ECMO, using Continuous Evaluation of VV-ECMO Outcomes (CEVVO), a deep learning-based model for predicting success of decannulation in patients supported on VV-ECMO. The running metric may be applied daily to categorize patients into high-risk and low-risk groups. Using these data, providers may consider initiating a weaning trial based on their expertise and CEVVO. Data were collected from 118 patients supported with VV-ECMO at the Columbia University Irving Medical Center. Using a long short-term memory-based network, CEVVO is the first model capable of integrating discrete clinical information with continuous data collected from an ECMO device. A total of 12 sets of 5-fold cross validations were conducted to assess the performance, which was measured using the area under the receiver operating characteristic curve (AUROC) and average precision (AP). To translate the predicted values into a clinically useful metric, the model results were calibrated and stratified into risk groups, ranging from 0 (high risk) to 3 (low risk). To further investigate the performance edge of CEVVO, 2 synthetic data sets were generated using Gaussian process regression. The first data set preserved the long-term dependency of the patient data set, whereas the second did not. CEVVO demonstrated consistently superior classification performance compared with contemporary models (P<.001 and P=.04 compared with the next highest AUROC and AP). Although the model's patient-by-patient predictive power may be too low to be integrated into a clinical setting (AUROC 95% CI 0.6822-0.7055; AP 95% CI 0.8515-0.8682), the patient risk classification system displayed greater potential. When measured at 72 hours, the high-risk group had a successful decannulation rate of 58% (7/12), whereas the low-risk group had a successful decannulation rate of 92% (11/12; P=.04). When measured at 96 hours, the high- and low-risk groups had a successful decannulation rate of 54% (6/11) and 100% (9/9), respectively (P=.01). We hypothesized that the improved performance of CEVVO was owing to its ability to efficiently capture transient temporal patterns. Indeed, CEVVO exhibited improved performance on synthetic data with inherent temporal dependencies (P<.001) compared with logistic regression and a dense neural network. The ability to interpret and integrate large data sets is paramount for creating accurate models capable of assisting clinicians in risk stratifying patients supported on VV-ECMO. Our framework may guide future incorporation of CEVVO into more comprehensive intensive care monitoring systems.

Spatial–Temporal Feature Network for Speech-Based Depression Recognition

Depression is a serious mental disorder that has received increased attention from society. Due to the advantage of easy acquisition of speech, researchers have tried to propose various automatic depression recognition algorithms based on speech. Feature selection and algorithm design are the main difficulties in speech-based depression recognition. In our work, we propose the spatial-temporal feature network (STFN) for depression recognition, which can capture the long-term temporal dependence of audio sequences. First, to obtain a better feature representation for depression analysis, we develop a self-supervised learning framework, called Vector Quantized Wav2vec Transformer Net (VQWTNet) to map speech features and phonemes with transfer learning. Second, the stacked gated residual blocks in SFNet are used in the model to integrate causal and dilated convolutions to capture multiscale contextual information by continuously expanding the receptive field. In addition, instead of LSTM, our method employs the hierarchical contrastive predictive coding (HCPC) loss in HCPCNet to capture the long-term temporal dependencies of speech, reducing the number of parameters while making the network easier to train. Finally, experimental results on DAIC-WOZ (AVEC 2017) and E-DAIC (AVEC 2019) show that the proposed model significantly improves the accuracy of depression recognition. On both datasets, the performance of our method far exceeds the baseline and achieves competitive results compared to state-of-the-art methods.

The predictive value of anticholinergic drug exposure and the outcome of pneumonia: a Danish database study.

Older adults are susceptible to anticholinergic effects. Dysphagia and pneumonia are associated with anticholinergic usage, though a definitive causative relationship has not been established. There is no effective way to predict the prognosis of older adults with pneumonia; therefore, this study investigates the predictive value of anticholinergic burden. Patients aged 65years and above admitted for community-acquired pneumonia from 2011 to 2018 in Denmark were included through Danish registries. We calculated anticholinergic drug exposure using the CRIDECO Anticholinergic Load Scale (CALS). The primary outcome was in-hospital mortality, and other outcomes included intensive care unit admission, ventilator usage, length of stay, 30-day/90-day/1-year mortality, institutionalisation, home care utilisation and readmission. 186,735 patients were included in the in-hospital outcome analyses, 165,181 in the readmission analysis, 150,791 in the institutionalisation analysis, and 95,197 and 73,461 patients in the home care analysis at follow-up. Higher CALS score was associated with higher in-hospital mortality, with a mean risk increasing from 9.9% (CALS 0) to 16.4% (CALS >10), though the risk plateaued above a CALS score of 8. A higher CALS score was also associated with greater mortality after discharge, more home health care, more institutionalizations and higher readmission rates. High anticholinergic burden levels were associated with poor patient outcomes including short-/long-term mortality, dependence and readmission. It may be useful to calculate the CALS score on admission of older patients with pneumonia to predict their prognosis. This also highlights the importance of avoiding the use of drugs with a high anticholinergic burden in older patients.

Physical informed memory networks for solving PDEs: implementation and applications

With the advent of physics informed neural networks (PINNs), deep learning has gained interest for solving nonlinear partial differential equations (PDEs) in recent years. In this paper, physics informed memory networks (PIMNs) are proposed as a new approach to solving PDEs by using physical laws and dynamic behavior of PDEs. Unlike the fully connected structure of the PINNs, the PIMNs construct the long-term dependence of the dynamics behavior with the help of the long short-term memory network. Meanwhile, the PDEs residuals are approximated using difference schemes in the form of convolution filter, which avoids information loss at the neighborhood of the sampling points. Finally, the performance of the PIMNs is assessed by solving the KdV equation and the nonlinear Schrödinger equation, and the effects of difference schemes, boundary conditions, network structure and mesh size on the solutions are discussed. Experiments show that the PIMNs are insensitive to boundary conditions and have excellent solution accuracy even with only the initial conditions.

Safety of the immediate use of nonsteroidal anti-inflammatory drugs after adult craniotomy for tumor.

Poor pain control has a negative impact on postoperative recovery and patient satisfaction. However, overzealous pain management, particularly with opioids, can confound serial neurological assessments, increase morbidity, and predispose patients to long-term dependence. Nonsteroidal anti-inflammatory drugs (NSAIDs) are effective in treating postoperative pain and can limit opioid intake, but their use has been limited in patients undergoing craniotomy for brain tumor resection due to concerns of an increased hemorrhage risk. Herein, the authors aim to 1) address the safety of NSAID use in the immediate postoperative setting and 2) determine whether NSAID administration decreases opioid use following craniotomy for tumor resection in adult patients. The authors conducted a retrospective cohort study of patients 18 years and older with an estimated glomerular filtration rate ≥ 60 ml/min/body surface area who had undergone craniotomy for tumor resection at their institution between 2019 and 2021. NSAID use in the first 48 hours following surgery was recorded. Primary outcomes were postoperative hemorrhage requiring a return to the operating room before hospital discharge and within 30 days of surgery. Secondary outcomes were more-than-minimal hemorrhage that did not require reoperation, acute kidney injury, and total opioid use within 48 hours after craniotomy. Among 1765 reviewed patient records, 1182 were eligible for inclusion in this analysis. Amid these records were 114 patients (9.6%) who had received at least one dose of an NSAID within 48 hours of their craniotomy. Four (0.3%) patients experienced bleeding requiring a return to operating room, one of whom was from the NSAID-treated group (RR 3.12, 95% CI 0.33-29.77, p = 0.30). No significant difference in nonoperative intracranial hemorrhage (RR 1.34, 95% CI 0.54-3.35, p = 0.53), postoperative acute kidney injury, or clinically significant extracranial bleeding was found between the NSAID and no-NSAID groups. Patients in the NSAID group had significantly higher oral morphine equivalent use (median 68 vs 30, p < 0.001). Postoperative NSAID use following craniotomy for tumor resection was not associated with an increased risk of hemorrhage requiring a return to the operating room. The authors noted higher opioid use in the patients treated with NSAIDs, which may reflect underlying reasons for the decision to treat patients with NSAIDs in the immediate postoperative period. These data warrant further investigation of NSAIDs as a safe, opioid-sparing postoperative pain management strategy in patients with normal kidney function who are undergoing intracranial tumor resection.

Gated Recursive and Sequential Deep Hierarchical Encoding for Detecting Incongruent News Articles

With the increase in misinformation across digital platforms, incongruent news detection is becoming an important research problem. Earlier, researchers have exploited various feature engineering approaches and deep learning models with embedding to capture incongruity between news headlines and their respective bodies. Studies have broadly considered different combinations of bag-of-words-based features, sequential encoding, hierarchical encoding, headline-guided attention-based encoding, and so on, of the text in headlines and bodies. In this article, we focus on addressing two important limitations observed with hierarchical encoding and headline-guided attention-based encoding methods. The existing hierarchical encoding-based studies limit the hierarchical structure of the body of a news article to paragraph level only, undermining the importance of incorporating long-term dependence from word level to sentence, paragraph, and body. Furthermore, the existing headline-guided attention-based encoding focuses on contextually similar contents in the body of the headline, undermining the importance of incorporating contextually dissimilar contents. Motivated by the above observations, this article proposes a gated recursive and sequential deep hierarchical encoding (GraSHE) method for detecting incongruent news articles by extending the hierarchical structure of the news body from the body to the word level and incorporating incongruity weight. From various experimental setups over three publicly available benchmarks datasets, the experimental results indicate that the proposed model outperforms baseline models with bag-of-word-based features, sequential, hierarchical, and headline-guided attention-based encoding methods. To further validate the performance of the proposed model, we conduct several ablation studies. The following key observations can be made from the ablation study: 1) models with hierarchical encoding outperform models with nonhierarchical encoding; 2) recursive encoding of sentences boosts the performance of models as compared with sequential encoding of sentences within paragraphs; and 3) incongruent news article detection is domain-dependent. Incorporating explicit features further boosts the performance of proposed model and also decreases the domain dependence of models.

Optimization and research of smart grid load forecasting model based on deep learning

Abstract In order to accurately grasp the power consumption of the market and deeply understand the operation law of the power market, the power grid must achieve accurate power load prediction by studying the supply and demand of the power system and the change law. Therefore, a smart grid load prediction analysis method based on deep learning is proposed. Firstly, DenseNet network is used to extract features, so as to achieve feature reuse and improve efficiency. Then, combining the long short-term memory (LSTM) network based on attention mechanism, a combined model DenseNet–AM–LSTM is proposed to solve the long-term dependence problem of feature extraction. After using batch normalization, the model can use higher learning rate to improve the training speed of the model and obtain the best prediction accuracy of the model. The experimental results show that this method can accurately predict the power load, and the accuracy is higher than the current popular methods, and it is expected to become an important technical support for solving the core problems of smart grid.

Stock Price Prediction Using Sentiment Analysis

Abstract: This research discusses the ever-changing nature of the stock market by using models such as LSTM and ARIMA to forecast stock prices. It recognizes the volatility and unpredictability of financial markets recognizing the need for robust tools that can accurately predict. Sentiment analysis is used for enhancing the accuracy of stock price prediction, drawing information from various sources like news resources and social media platforms that reflect public sentiments and opinions. We aim to analyze a thorough aspect of factors influencing stock prices by combining these sentiments with the models. The LSTM model is used to study long-term dependencies in stock price trends and ARIMA provides information into the time series components. Combining sentiment analysis helps us to scan the emotional tone around a particular stock and contributes valuable data to the prediction process. The integration between the machine learning models and sentiment analysis offers an extensive approach to predicting stock prices considering both past trends and current public sentiment. This study adds to the work in progress to develop more accurate and adaptive tools for exploring the difficulties of the stock market.

Discovering Process Models with Long-Term Dependencies while Providing Guarantees and Filtering Infrequent Behavior Patterns

In process discovery, the goal is to find, for a given event log, the model describing the underlying process. While process models can be represented in a variety of ways, Petri nets form a theoretically well-explored description language and are therefore often used. In this paper, we extend the eST-Miner process discovery algorithm. The eST-Miner computes a set of Petri net places which are considered to be fitting with respect to a certain fraction of the behavior described by the given event log as indicated by a given noise threshold. It evaluates all possible candidate places using token-based replay. The set of replayable traces is determined for each place in isolation, i.e., these sets do not need to be consistent. This allows the algorithm to abstract from infrequent behavioral patterns occurring only in some traces. However, when combining places into a Petri net by connecting them to the corresponding uniquely labeled transitions, the resulting net can replay exactly those traces from the event log that are allowed by the combination of all inserted places. Thus, inserting places one-by-one without considering their combined effect may result in deadlocks and low fitness of the Petri net. In this paper, we explore adaptions of the eST-Miner, that aim to select a subset of places such that the resulting Petri net guarantees a definable minimal fitness while maintaining high precision with respect to the input event log. Furthermore, current place evaluation techniques tend to block the execution of infrequent activity labels. Thus, a refined place fitness metric is introduced and thoroughly investigated. In our experiments we use real and artificial event logs to evaluate and compare the impact of the various place selection strategies and place fitness evaluation metrics on the returned Petri net.