Browsing by Subject "Algoritmit"

Sort by: Order: Results:

Now showing items 1-20 of 41
  • Jabs, Christoph (Helsingin yliopisto, 2022)
    Many real-world problem settings give rise to NP-hard combinatorial optimization problems. This results in a need for non-trivial algorithmic approaches for finding optimal solutions to such problems. Many such approaches—ranging from probabilistic and meta-heuristic algorithms to declarative programming—have been presented for optimization problems with a single objective. Less work has been done on approaches for optimization problems with multiple objectives. We present BiOptSat, an exact declarative approach for finding so-called Pareto-optimal solutions to bi-objective optimization problems. A bi-objective optimization problem arises for example when learning interpretable classifiers and the size, as well as the classification error of the classifier should be taken into account as objectives. Using propositional logic as a declarative programming language, we seek to extend the progress and success in maximum satisfiability (MaxSAT) solving to two objectives. BiOptSat can be viewed as an instantiation of the lexicographic method and makes use of a single SAT solver that is preserved throughout the entire search procedure. It allows for solving three tasks for bi-objective optimization: finding a single Pareto-optimal solution, finding one representative solution for each Pareto point, and enumerating all Pareto-optimal solutions. We provide an open-source implementation of five variants of BiOptSat, building on different algorithms proposed for MaxSAT. Additionally, we empirically evaluate these five variants, comparing their runtime performance to that of three key competing algorithmic approaches. The empirical comparison in the contexts of learning interpretable decision rules and bi-objective set covering shows practical benefits of our approach. Furthermore, for the best-performing variant of BiOptSat, we study the effects of proposed refinements to determine their effectiveness.
  • Avikainen, Jari (Helsingin yliopisto, 2019)
    This thesis presents a wavelet-based method for detecting moments of fast change in the textual contents of historical newspapers. The method works by generating time series of the relative frequencies of different words in the newspaper contents over time, and calculating their wavelet transforms. Wavelet transform is essentially a group of transformations describing the changes happening in the original time series at different time scales, and can therefore be used to pinpoint moments of fast change in the data. The produced wavelet transforms are then used to detect fast changes in word frequencies by examining products of multiple scales of the transform. The properties of the wavelet transform and the related multi-scale product are evaluated in relation to detecting various kinds of steps and spikes in different noise environments. The suitability of the method for analysing historical newspaper archives is examined using an example corpus consisting of 487 issues of Uusi Suometar from 1869–1918 and 250 issues of Wiipuri from 1893–1918. Two problematic features in the newspaper data, noise caused by OCR (optical character recognition) errors and uneven temporal distribution of the data, are identified and their effects on the results of the presented method are evaluated using synthetic data. Finally, the method is tested using the example corpus, and the results are examined briefly. The method is found to be adversely affected especially by the uneven temporal distribution of the newspaper data. Without additional processing, or improving the quality of the examined data, a significant amount of the detected steps are due to the noise in the data. Various ways of alleviating the effect are proposed, among other suggested improvements on the system.
  • Wargelin, Matias (Helsingin yliopisto, 2021)
    Musical pattern discovery refers to the automated discovery of important repeated patterns, such as melodies and themes, from music data. Several algorithms have been developed to solve this problem, but evaluating the algorithms has been difficult without proper visualisations of the output of the algorithms. To address this issue a web application named Mupadie was built. Mupadie accepts MIDI music files as input and visualises the outputs of musical pattern discovery algorithms, with implementations of SIATEC and TTWIA built in the application. Other algorithms can be visualised if the algorithm output is uploaded to Mupadie as a JSON file that follows a specified data structure. Using Mupadie, an evaluation of SIATEC and TTWIA was conducted. Mupadie was found to be a useful tool in the qualitative evaluation of these musical pattern discovery algorithms; it helped reveal systematically recurring issues with the discovered patterns, some previously known and some previously undocumented. The findings were then used to suggest improvements to the algorithms.
  • Lehtonen, Leevi (Helsingin yliopisto, 2021)
    Quantum computing has an enormous potential in machine learning, where problems can quickly scale to be intractable for classical computation. A Boltzmann machine is a well-known energy-based graphical model suitable for various machine learning tasks. Plenty of work has already been conducted for realizing Boltzmann machines in quantum computing, all of which have somewhat different characteristics. In this thesis, we conduct a survey of the state-of-the-art in quantum Boltzmann machines and their training approaches. Primarily, we examine variational quantum Boltzmann machine, a specific variant of quantum Boltzmann machine suitable for the near-term quantum hardware. Moreover, as variational quantum Boltzmann machine heavily relies on variational quantum imaginary time evolution, we effectively analyze variational quantum imaginary time evolution to a great extent. Compared to the previous work, we evaluate the execution of variational quantum imaginary time evolution with a more comprehensive collection of hyperparameters. Furthermore, we train variational quantum Boltzmann machines using a toy problem of bars and stripes, representing more multimodal probability distribution than the Bell states and the Greenberger-Horne-Zeilinger states considered in the earlier studies.
  • Harviainen, Juha (Helsingin yliopisto, 2021)
    Computing the permanent of a matrix is a famous #P-hard problem with a wide range of applications. The fastest known exact algorithms for the problem require an exponential number of operations, and all known fully polynomial randomized approximation schemes are rather complicated to implement and have impractical time complexities. The most promising recent advancements on approximating the permanent are based on rejection sampling and upper bounds for the permanent. In this thesis, we improve the current state of the art by developing the deep rejection sampling method, which combines an exact algorithm with the rejection sampling method. The algorithm precomputes a dynamic programming table that tightens the initial upper bound used by the rejection sampling method. In a sense, the table is used to jump-start the sampling process. We give a high probability upper bound for the time complexity of the deep rejection sampling method for random (0, 1)-matrices in which each entry is 1 with probability p. For matrices with p < 1/5, our high probability bound is stronger than in previous work. In addition to that, we empirically observe that our algorithm outperforms earlier rejection sampling methods by testing it with different parameters against other algorithms on multiple classes of matrices. The improvements in sampling times are especially notable in cases in which the ratios of the permanental upper bounds and the exact value of the permanent are huge.
  • Liu, Yang Jr (Helsingin yliopisto, 2020)
    Automatic readability assessment is considered as a challenging task in NLP due to its high degree of subjectivity. The majority prior work in assessing readability has focused on identifying the level of education necessary for comprehension without the consideration of text quality, i.e., how naturally the text flows from the perspective of a native speaker. Therefore, in this thesis, we aim to use language models, trained on well-written prose, to measure not only text readability in terms of comprehension but text quality. In this thesis, we developed two word-level metrics based on the concordance of article text with predictions made using language models to assess text readability and quality. We evaluate both metrics on a set of corpora used for readability assessment or automated essay scoring (AES) by measuring the correlation between scores assigned by our metrics and human raters. According to the experimental results, our metrics are strongly correlated with text quality, which achieve 0.4-0.6 correlations on 7 out of 9 datasets. We demonstrate that GPT-2 surpasses other language models, including the bigram model, LSTM, and bidirectional LSTM, on the task of estimating text quality in a zero-shot setting, and GPT-2 perplexity-based measure is a reasonable indicator for text quality evaluation.
  • Liu, Yang (Helsingin yliopisto, 2020)
    Automatic readability assessment is considered as a challenging task in NLP due to its high degree of subjectivity. The majority prior work in assessing readability has focused on identifying the level of education necessary for comprehension without the consideration of text quality, i.e., how naturally the text flows from the perspective of a native speaker. Therefore, in this thesis, we aim to use language models, trained on well-written prose, to measure not only text readability in terms of comprehension but text quality. In this thesis, we developed two word-level metrics based on the concordance of article text with predictions made using language models to assess text readability and quality. We evaluate both metrics on a set of corpora used for readability assessment or automated essay scoring (AES) by measuring the correlation between scores assigned by our metrics and human raters. According to the experimental results, our metrics are strongly correlated with text quality, which achieve 0.4-0.6 correlations on 7 out of 9 datasets. We demonstrate that GPT-2 surpasses other language models, including the bigram model, LSTM, and bidirectional LSTM, on the task of estimating text quality in a zero-shot setting, and GPT-2 perplexity-based measure is a reasonable indicator for text quality evaluation.
  • Mylläri, Juha (Helsingin yliopisto, 2022)
    Anomaly detection in images is the machine learning task of classifying inputs as normal or anomalous. Anomaly localization is the related task of segmenting input images into normal and anomalous regions. The output of an anomaly localization model is a 2D array, called an anomaly map, of pixel-level anomaly scores. For example, an anomaly localization model trained on images of non-defective industrial products should output high anomaly scores in image regions corresponding to visible defects. In unsupervised anomaly localization the model is trained solely on normal data, i.e. without labelled training observations that contain anomalies. This is often necessary as anomalous observations may be hard to obtain in sufficient quantities and labelling them is time-consuming and costly. Student-teacher feature pyramid matching (STFPM) is a recent and powerful method for unsupervised anomaly detection and localization that uses a pair of convolutional neural networks of identical architecture. In this thesis we propose two methods of augmenting STFPM to produce better segmentations. Our first method, discrepancy scaling, significantly improves the segmentation performance of STFPM by leveraging pre-calculated statistics containing information about the model’s behaviour on normal data. Our second method, student-teacher model assisted segmentation, uses a frozen STFPM model as a feature detector for a segmentation model which is then trained on data with artificially generated anomalies. Using this second method we are able to produce sharper anomaly maps for which it is easier to set a threshold value that produces good segmentations. Finally, we propose the concept of expected goodness of segmentation, a way of assessing the performance of unsupervised anomaly localization models that, in contrast to current metrics, explicitly takes into account the fact that a segmentation threshold needs to be set. Our primary method, discrepancy scaling, improves segmentation AUROC on the MVTec AD dataset over the base model by 13%, measured in the shrinkage of the residual (1.0 − AUROC). On the image-level anomaly detection task, a variant of the discrepancy scaling method improves performance by 12%.
  • Thapa Magar, Purushottam (Helsingin yliopisto, 2021)
    Rapid growth and advancement of next generation sequencing (NGS) technologies have changed the landscape of genomic medicine. Today, clinical laboratories perform DNA sequencing on a regular basis, which is an error prone process. Erroneous data affects downstream analysis and produces fallacious result. Therefore, external quality assessment (EQA) of laboratories working with NGS data is crucial. Validation of variations such as single nucleotide polymor- phism (SNP) and InDels (<50 bp) is fairly accurate these days. However, detection and quality assessment of large changes such as the copy number variation (CNV) continues to be a concern. In this work, we aimed to study the feasibility of an automated CNV concordance analysis for the laboratory EQA services. We benchmarked variants reported by 25 laboratories against the highly curated gold standard for the son (HG002/NA24385) of the askenazim trio from the Personal Genome Project published by the Genome in a Bottle Consortium (GIAB). We employed two methods to conduct concordance of CNVs, the sequence based comparison with Truvari and the in-house exome-based comparison. For deletion calls of two whole genome sequencing (WGS) submissions, Truvari gained a value greater than 88% and 68% for precision and recall respectively. Conversely, the in-house method’s precision and recall score peaked at 39% and 7.9% respectively for one WGS submission for both deletion and duplication calls. The results indicate that automated CNV concordance analysis of the deletion calls for the WGS-based callset might be feasible with Truvari. On the other hand, results for panel-based targeted sequencing for the deletion calls showed precision and recall rates ranging from 0-80% and 0-5.6% respectively with Truvari. The result suggests that automated concordance analysis of CNVs for targeted sequencing remains a challenge. In conclusion, CNV concordance analysis depends on how the sequence data is generated.
  • Gold, Ayoola (Helsingin yliopisto, 2021)
    The importance of Automatic Speech Recognition cannot be underestimated in today’s worlds as they play a significant role in human computer interaction. ASR systems have been studied deeply over time, but their maximum potential is yet to be explored for the Finnish language. Development of a traditional ASR system involves a number of hand-crafted engineering which has made this technology quite difficult and resourceful to develop. However, with advancements in the field of neural networks, end-to-end ASR neural networks can be developed which can automatically learn the mappings of audio to its corresponding transcript., therefore reducing hand crafted engineering requirements. End-to-end neural network ASR systems have been largely developed commercially by tech giants such as Microsoft, Google and Amazon. However, there are limitations to these commercial services such as data privacy and cost of usage. In this thesis, we explored existing studies in the development of an end-to-end neural network ASR for Finnish language. One successful technique utilized in the development of neural network ASR in the advent of inadequate data is Transfer learning. This is the approach explored in this thesis for the development of the end-to-end neural network ASR system. In addition, the success of this approach was evaluated. In order to achieve this purpose, dataset collected from the Finnish Bank of Finland and Kaggle were used to fine-tune Mozilla DeepSpeech model which is a pretrained end-to-end neural network ASR in English language. The results obtained by fine-tuning the pretrained neural network ASR in English for Finnish language showed a word error rate as low as 40% and character error rate as low as 22%. We therefore concluded that transfer learning is a successful technique for creating ASR model for a new language using a pretrained model in another language with little effort, data and resources.
  • Kinnunen, Lauri (Helsingin yliopisto, 2022)
    This thesis is a review of articles focusing on software assisted floor plan design for architecture. I group the articles into optimization, case based design, and machine learning, based on their use of prior examples. I then look into each category and further classify articles based on dimensions relevant to their overall approach. Case based design was a popular research field in the 1990s and early 2000s when several large research projects were conducted. However, since then the research has slowed down. Over the past 20 years, optimization methods to solve architectural floor plans have been researched extensively using a number of different algorithms and data models. The most popular approach is to use a stochastic optimization method such as a genetic algorithm or simulated annealing. More recently, a number of articles have investigated the possibility of using machine learning on architectural floor plans. The advent of neural networks and GAN models, in particular, has spurred a great deal of new research. Despite considerable research efforts, assisted floor plan design has not found its way into commercial applications. To aid industry adoption, more work is needed on the integration of computational design tools into the existing design workflows.
  • Sarapisto, Teemu (Helsingin yliopisto, 2022)
    In this thesis we investigate the feasibility of machine learning methods for estimating the type and the weight of individual food items from images taken of customers’ plates at a buffet- style restaurant. The images were collected in collaboration with the University of Turku and Flavoria, a public lunch-line restaurant, where a camera was mounted above the cashier to automatically take a photo of the foods chosen by the customer when they went to pay. For each image, an existing system of scales at the restaurant provided the weights for each individual food item. We describe suitable model architectures and training setups for the weight estimation and food identification tasks and explain the models’ theoretical background. Furthermore we propose and compare two methods for utilizing a restaurant’s daily menu information for improving model performance in both tasks. We show that the models perform well in comparison to baseline methods and reach accuracy on par with other similar work. Additionally, as the images were captured automatically, in some of the images the food was occluded or blurry, or the image contained sensitive customer information. To address this we present computer vision techniques for preprocessing and filtering the images. We publish the dataset containing the preprocessed images along with the corresponding individual food weights for use in future research. The main results of the project have been published as a peer-reviewed article in the International Conference in Pattern Recognition Systems 2022. The article received the best paper award of the conference.
  • Porttinen, Peter (Helsingin yliopisto, 2020)
    Computing an edit distance between strings is one of the central problems in both string processing and bioinformatics. Optimal solutions to edit distance are quadratic to the lengths of the input strings. The goal of this thesis is to study a new approach to approximate edit distance. We use a chaining algorithm presented by Mäkinen and Sahlin in "Chaining with overlaps revisited" CPM 2020 implemented verbatim. Building on the chaining algorithm, our focus is on efficiently finding a good set of anchors for the chaining algorithm. We present three approaches to computing the anchors as maximal exact matches: Bi-Directional Burrows-Wheeler Transform, Minimizers, and lastly, a hybrid implementation of the two. Using the maximal exact matches as anchors, we can efficiently compute an optimal chaining alignment for the strings. The chaining alignment further allows us to determine all such intervals where mismatches occur by looking at which sequences are not in the chain. Using these smaller intervals lets us approximate edit distance with a high degree of accuracy and a significant speed improvement. The methods described present a way to approximate edit distance in time complexity bounded by the number of maximal exact matches.
  • Ma, Jun (Helsingin yliopisto, 2021)
    Sequence alignment by exact or approximate string matching is one of the fundamental problems in bioinformatics. As the volume of sequenced genomes grows rapidly, pairwise sequence alignment becomes inefficient for pan-genomic analyses involving multiple sequences. The graph representation of multiple genomes has been an increasingly useful tool in pan-genomics research. Therefore, sequence-to-graph alignment becomes an important and challenging problem. For pairwise approximate sequence alignment under Levenshtein (edit) distance, subquadratic algorithms for finding an optimal solution are unknown. As a result, aligning sequences of millions of characters optimally is too challenging and impractical. Thus, many heuristics and techniques are developed for possibly suboptimal alignments. Among them, co-linear chaining (CLC) is a powerful and popular technique that approximates the alignment by finding a chain of short aligned fragments that may come from exact matching. The optimal solution to CLC on sequences can be found efficiently in subquadratic time. For sequence-to-graph alignment, the CLC problem has been solved theoretically on a special class of graphs that are narrow and have no cycles, i.e. directed acyclic graphs (DAGs) with small width, by Mäkinen et al. (ACM Transactions on Algorithms, 2019). Pan-genome graphs such as variation graphs satisfy these restrictions but allowing cycles may enable more general applications of the algorithm. In this thesis, we introduce an efficient algorithm to solve the CLC problem on general graphs with small width that may have cycles, by reducing it to a slightly modified CLC problem on DAGs. We implemented an initial version of the new algorithm on DAGs as a sequence-to-graph aligner GraphChainer. The aligner is evaluated and compared to an existing state-of-the-art aligner GraphAligner (Genome Biology, 2020) in experiments using both simulated and real genome assembly data on variation graphs. Our method improves the quality of alignments significantly in the task of aligning real human PacBio data. GraphChainer is freely available as an open source tool at https://github.com/algbio/GraphChainer.
  • Länsman, Olá-Mihkku (Helsingin yliopisto, 2020)
    Demand forecasts are required for optimizing multiple challenges in the retail industry, and they can be used to reduce spoilage and excess inventory sizes. The classical forecasting methods provide point forecasts and do not quantify the uncertainty of the process. We evaluate multiple predictive posterior approximation methods with a Bayesian generalized linear model that captures weekly and yearly seasonality, changing trends and promotional effects. The model uses negative binomial as the sampling distribution because of the ability to scale the variance as a quadratic function of the mean. The forecasting methods provide highest posterior density intervals in different credible levels ranging from 50% to 95%. They are evaluated with proper scoring function and calculation of hit rates. We also measure the duration of the calculations as an important result due to the scalability requirements of the retail industry. The forecasting methods are Laplace approximation, Monte Carlo Markov Chain method, Automatic Differentiation Variational Inference, and maximum a posteriori inference. Our results show that the Markov Chain Monte Carlo method is too slow for practical use, while the rest of the approximation methods can be considered for practical use. We found out that Laplace approximation and Automatic Differentiation Variational Inference have results closer to the method with best analytical quarantees, the Markov Chain Monte Carlo method, suggesting that they were better approximations of the model. The model faced difficulties with highly promotional, slow selling, and intermittent data. Best fit was provided with high selling SKUs, for which the model provided intervals with hit rates that matched the levels of the credible intervals.
  • Laitala, Julius (Helsingin yliopisto, 2021)
    Arranging products in stores according to planograms, optimized product arrangement maps, is important for keeping up with the highly competitive modern retail market. The planograms are realized into product arrangements by humans, a process which is prone to mistakes. Therefore, for optimal merchandising performance, the planogram compliance of the arrangements needs to be evaluated from time to time. We investigate utilizing a computer vision problem setting – retail product detection – to automate planogram compliance evaluation. We introduce the relevant problems, the state-of- the-art approaches for solving them and background information necessary for understanding them. We then propose a computer vision based planogram compliance evaluation pipeline based on the current state of the art. We build our proposed models and algorithms using PyTorch, and run tests against public datasets and an internal dataset collected from a large Nordic retailer. We find that while the retail product detection performance of our proposed approach is quite good, the planogram compliance evaluation performance of our whole pipeline leaves a lot of room for improvement. Still, our approach seems promising, and we propose multiple ways for improving the performance enough to enable possible real world utility. The code used for our experiments and the weights for our models are available at https://github.com/laitalaj/cvpce
  • Fred, Hilla (Helsingin yliopisto, 2022)
    Improving the monitoring of health and well-being of dairy cows through the use of computer vision based systems is a topic of ongoing research. A reliable and low-cost method for identifying cow individuals would enable automatic detection of stress, sickness or injury, and the daily observation of the animals would be made easier. Neural networks have been used successfully in the identification of cow individuals, but methods are needed that do not require incessant annotation work to generate training datasets when there are changes within a group. Methods for person re-identification and tracking have been researched extensively, with the aim of generalizing beyond the training set. These methods have been found suitable also for re-identifying and tracking previously unseen dairy cows in video frames. In this thesis, a metric-learning based re-identification model pre-trained on an existing cow dataset is compared to a similar model that has been trained on new video data recorded at Luke Maaninka research farm in Spring 2021, which contains 24 individually labelled cow individuals. The models are evaluated in tracking context as appearance descriptors in Kalman filter based tracking algorithm. The test data is video footage from a separate enclosure in Maaninka and a group of 24 previously unseen cow individuals. In addition, a simple procedure is proposed for the automatic labeling of cow identities in images based on RFID data collected from cow ear tags and feeding stations, and the known feeding station locations.
  • Hertweck, Corinna (Helsingin yliopisto, 2020)
    In this work, we seek robust methods for designing affirmative action policies for university admissions. Specifically, we study university admissions under a real centralized system that uses grades and standardized test scores to match applicants to university programs. For the purposes of affirmative action, we consider policies that assign bonus points to applicants from underrepresented groups with the goal of preventing large gaps in admission rates across groups, while ensuring that the admitted students are for the most part those with the highest scores. Since such policies have to be announced before the start of the application period, there is uncertainty about which students will apply to which programs. This poses a difficult challenge for policy-makers. Hence, we introduce a strategy to design policies for the upcoming round of applications that can either address a single or multiple demographic groups. Our strategy is based on application data from previous years and a predictive model trained on this data. By comparing this predictive strategy to simpler strategies based only on application data from, e.g., the previous year, we show that the predictive strategy is generally more conservative in its policy suggestions. As a result, policies suggested by the predictive strategy lead to more robust effects and fewer cases where the gap in admission rates is inadvertently increased through the suggested policy intervention. Our findings imply that universities can employ predictive methods to increase the reliability of the effects expected from the implementation of an affirmative action policy.
  • Ikkala, Tapio (Helsingin yliopisto, 2020)
    This thesis presents a scalable method for identifying anomalous periods of non-activity in short periodic event sequences. The method is tested with real world point-of-sale (POS) data from grocery retail setting. However, the method can be applied also to other problem domains which produce similar sequential data. The proposed method models the underlying event sequence as a non-homogeneous Poisson process with a piecewise constant rate function. The rate function for the piecewise homogeneous Poisson process can be estimated with a change point detection algorithm that minimises a cost function consisting of the negative Poisson log-likelihood and a penalty term that is linear to the number of change points. The resulting model can be queried for anomalously long periods of time with no events, i.e., waiting times, by defining a threshold below which the waiting time observations are deemed anomalies. The first experimental part of the thesis focuses on model selection, i.e., in finding a penalty value that results in the change point detection algorithm detecting the true changes in the intensity of the arrivals of the events while not reacting to random fluctuations in the data. In the second experimental part the performance of the anomaly detection methodology is measured against stock-out data, which gives an approximate ground truth for the termination of a POS event sequence. The performance of the anomaly detector is found to be subpar in terms of precision and recall, i.e., the true positive rate and the positive predictive value. The number of false positives remains high even with small threshold values. This needs to be taken into account when considering applying the anomaly detection procedure in practice. Nevertheless, the methodology may have practical value in the retail setting, e.g., in guiding the store personnel where to focus their resources in ensuring the availability of the products.
  • Duong, Quoc Quan (Helsingin yliopisto, 2021)
    Discourse dynamics is one of the important fields in digital humanities research. Over time, the perspectives and concerns of society on particular topics or events might change. Based on the changing in popularity of a certain theme different patterns are formed, increasing or decreasing the prominence of the theme in news. Tracking these changes is a challenging task. In a large text collection discourse themes are intertwined and uncategorized, which makes it hard to analyse them manually. The thesis tackles a novel task of automatic extraction of discourse trends from large text corpora. The main motivation for this work lies in the need in digital humanities to track discourse dynamics in diachronic corpora. Machine learning is a potential method to automate this task by learning patterns from the data. However, in many real use-cases ground truth is not available and annotating discourses on a corpus-level is incredibly difficult and time-consuming. This study proposes a novel procedure to generate synthetic datasets for this task, a quantitative evaluation method and a set of benchmarking models. Large-scale experiments are run using these synthetic datasets. The thesis demonstrates that a neural network model trained on such datasets can obtain meaningful results when applied to a real dataset, without any adjustments of the model.