Experiments and Results. The baseline MT systems (referred to as v0) were solely trained on out-of-domain data (parallel, monolingual, and development data from Europarl). First, we exploited the in-domain development data and used it in the first modification of the baseline system (v1) instead of the out-of-domain (Europarl) data. In this case, the individual system models (translation tables, language model, etc.) remained the same, but their importance (optimal weights in the Moses' log-linear framework) was different. The in-domain monolingual data could be exploited in two ways: a) to join the general domain data and the new in-domain data into one set, use it to train one language model and optimize its weight using MERT on the in-domain development data: b) to train a new separate language model from the new data, add it to the log-linear framework and let MERT optimize its weight together with other model weights. We tested both approaches. In system v2, we followed the first option (retraining the language model on an enlarged data) and in system v3, we followed the second option (training an additional language model and optimizing). An overview of the system versions trained for the first cycle evaluation is presented in Table 16. Version Parallel training data Monolingual training data Development data Test data v0 general general general in-domain v1 general general in-domain in-domain v2 general general + in-domain in-domain in-domain v3 general general | in-domain in-domain in-domain
Experiments and Results. We have two accurate methods to estimate time delays: i) EA-M-CV method is an evolutionary algorithm with mixed representation (integer and real numbers), and a objective function based on kernel formulation and cross-validation [12]. This method models a single underlying function that generates the two images plus the delay ∆ between them.
Experiments and Results. A. Experimental Setup We have evaluated the approach presented in Section III-B using two data sets, recorded in a static and a dynamic environment. The static data set is publicly available1. In the test data sets, the robot traverses a closed loop (see Fig. 10) multiple times with velocity of 1 m/s in an indoor environment. Both data sets were collected in the basement of O¨ rebro university using a commercial Automatically Guided Vehicle (AGV) system from Kollmorgen Automation AB. A 1Data sets are available under: xxxx://xxxxxx/xxxxxxxx.xxxx. Master Controller (VMC 5000) controls the vehicle along predefined trajectories. The ground truth was obtained with a commercial infrastructure-based positioning system, which tracks wall-mounted reflectors using a rotating laser. After setup and calibration, this system provides accurate (according to its specification accuracy should be approx. 1 cm or less) position information. For infra-structure free localisation we use a LIDAR with field of view of 270 degrees and range of 18 m. The data set covers a 25 m 25 m area. In both cases the robot was travelling along the same predefined path with the same velocity. To emulate a dynamic environment, we have asked a group of people to not only move around in the environment, but also on purpose to disturb the localisation process by changing the shape of the environment with panels or even to occlude the laser with them. The goal of the experiment was to investigate how using an informed prior (see Fig. 1) will affect global localisation. The comparison in this paper is done between uniform initialisa- tion of NDT-MCL and GMM initialisation of NDT-MCL, as described in Section III. We have chosen 60 random points along the path which rep- resent different starting positions for the localisation process. We use the following four evaluation criteria:
Experiments and Results. To compare the classification performance of our new representations with the simple representation of the previous chapter we have conducted the same experiments using the same settings and data sets (see Section 2.7). We will also compare our results to Ltree, OC1 and C4.5 and the other evolution- ary algorithms (esia and cefr-miner) already mentioned in the previous chapter. The tables with results also contain an extra column, labeled k, to indicate the number of clusters in the case of our clustering gp algorithms or the maximum number of partitions in the case of the gain gp and gain ratio gp algorithms. The best (average) result for each data set is printed in bold font. The entry N/A indicates that no results were available. To determine if the results obtained by our algorithms are statistically significantly different from the results reported for esia, cefr-miner, Ltree, OC1 and C4.5, we have performed two-tailed independent samples t-tests with a 95% confidence level (p = 0.05) using the reported mean and standard deviations. The null-hypothesis in each test is that the means of the two algorithms involved are equal. In order to determine whether the differences between our gp algorithms are statistically significant we used paired two- tailed t-tests with a 95% confidence level (p = 0.05) using the results of 100 runs (10 random seeds times 10 folds). In these tests the null-hypothesis is also that the means of the two algorithms involved are equal.
Experiments and Results. In order to assess the performance of our fuzzy representations we will com- pare them to our non-fuzzy representations from the previous chapters as well as with the other evolutionary (cefr-miner and esia) and non-evolutionary (Ltree, OC1 and C4.5) classification algorithms introduced in Chapter 2. N/A indicates that no results were available. The experimental setup and data sets are the same as in the previous chapters and are described in Sections 2.7 and 2.8. An overview of the most important gp parameters can be found in Table 4.1. In the case of our partitioning gp algorithms the criterion used, either gain or gain ratio, is indicated between brackets ‘(’ and ‘)’. The tables with results contain a column, labeled k, to indicate the number of clusters in the case of our clustering gp algorithms or the maximum number of parti- tions in the case of the partitioning gp algorithms. The best (average) result for each data set is printed in bold font. Because the partitioning gp algo- rithms do not scale well with parameter k we will only look at a maximum of three partitions, clusters or fuzzy sets per numerical valued attribute. Table 4.1: The main gp parameters. Parameter Value Population Size 100 Initialization ramped half-and-half Initial maximum tree depth 6 Maximum number of nodes 63 Parent Selection Tournament Selection Tournament Size 5 Evolutionary model (μ, λ) Offspring Size 200 Crossover Rate 0.9 Crossover Type swap subtree Mutation Rate 0.9 Mutation Type branch mutation Stop condition 99 generations To determine if the results obtained by our algorithms are statistically significantly different from the results reported for esia, cefr-miner, Ltree, OC1 and C4.5, we have performed two-tailed independent samples t-tests with a 95% confidence level (p = 0.05) using the reported mean and standard deviations. The null-hypothesis in each test is that the means of the two algorithms involved are equal. In order to determine whether the differences between our gp algorithms are statistically significant we used paired two- tailed t-tests with a 95% confidence level (p = 0.05) using the results of 100 runs (10 random seeds times 10 folds). In these tests the null-hypothesis is also that the means of the two algorithms involved are equal.
Experiments and Results. We used CrowdCrafting2 for recruiting workers because of a limited presence of Mongolian speakers on plat- forms such as Amazon Mechanical Turk and CrowdFlower. CrowdCrafting is free for scientific projects with volun- xxxx contributors. In phase 1, the total of 77 web users were asked to translate 947 manually built synsets from the space domain, that is, the subtree under the high-level synsets of space in (Ganbold et al., 2014b; Giunchiglia et al., 2009). In phase 2, 75 web users were asked to validate the results of phase 1. In total, contributors have completed 9,490 tasks and have introduced 6,442 words3. In order to evaluate contributions from the crowd, we com- piled a gold standard from the space domain in Mongolian, covering all synsets that were included in the crowdsourc- ing experiment. The gold standard corpus was created by 2xxxxx://xxxxxxxxxxxxx.xxx 3Data collected during the two phases are available at xxxxx://xxxxxxxxxxxxx.xxx/project/mongolian-lkc and at xxxxx://xxxxxxxxxxxxx.xxx/project/mongolian-lkc-evaluation under CC-BY-SA license.
