Helsinki Institute for Information Technology

About: Helsinki Institute for Information Technology is a facility organization based out in Espoo, Finland. It is known for research contribution in the topics: Population & Bayesian network. The organization has 630 authors who have published 1962 publications receiving 63426 citations.

Indexed Matching Statistics and Shortest Unique Substrings

Abstract: The unidirectional and bidirectional matching statistics between two strings s and t on alphabet Σ, and the shortest unique substrings of a single string t, are the cornerstone of a number of large-scale genome analysis applications, and they encode nontrivial structural properties of s and t. In this paper we compute for the first time the matching statistics between s and t in O((|s| + |t|)log|Σ|) time and in O(|s|log|Σ|) bits of space, circumventing the need for computing the depths of suffix tree nodes that characterized previous approaches. Symmetrically, we compute for the first time the shortest unique substrings of a string t in O(|t|log|Σ|) time and in O(|t|log|Σ|) bits of space. A key component of our methods is an encoding of both the unidirectional and the bidirectional statistics that takes 2|t| + o(|t|) bits of space and that allows constant-time access to every position.

New Error Bounds for Approximations from Projected Linear Equations

Abstract: We consider linear fixed point equations and their approximations by projection on a low dimensional subspace. We derive new bounds on the approximation error of the solution, which are expressed in terms of low dimensional matrices and can be computed by simulation. When the fixed point mapping is a contraction, as is typically the case in Markovian decision processes (MDP), one of our bounds is always sharper than the standard worst case bounds, and another one is often sharper. Our bounds also apply to the non-contraction case, including policy evaluation in MDP with nonstandard projections that enhance exploration. There are no error bounds currently available for this case to our knowledge.

Classifying Process Instances Using Recurrent Neural Networks

Abstract: Process Mining consists of techniques where logs created by operative systems are transformed into process models. In process mining tools it is often desired to be able to classify ongoing process instances, e.g., to predict how long the process will still require to complete, or to classify process instances to different classes based only on the activities that have occurred in the process instance thus far. Recurrent neural networks and its subclasses, such as Gated Recurrent Unit (GRU) and Long Short-Term Memory (LSTM), have been demonstrated to be able to learn relevant temporal features for subsequent classification tasks. In this paper we apply recurrent neural networks to classifying process instances. The proposed model is trained in a supervised fashion using labeled process instances extracted from event log traces. This is the first time we know of GRU having been used in classifying business process instances. Our main experimental results shows that GRU outperforms LSTM remarkably in training time while giving almost identical accuracies to LSTM models. Additional contributions of our paper are improving the classification model training time by filtering infrequent activities, which is a technique commonly used, e.g., in Natural Language Processing (NLP).

Manipulating Bodily Presence Affects Cross-Modal Spatial Attention: A Virtual-Reality-Based ERP Study

Abstract: Earlier studies have revealed cross-modal visuo-tactile interactions in endogenous spatial attention. The current research used event-related potentials (ERPs) and virtual reality (VR) to identify how the visual cues of the perceiver’s body affect visuo-tactile interaction in endogenous spatial attention and at what point in time the effect takes place. A bimodal oddball task with lateralized tactile and visual stimuli was presented in two VR conditions, one with and one without visible hands, and one VR-free control with hands in view. Participants were required to silently count one type of stimulus and ignore all other stimuli presented in irrelevant modality or location. The presence of hands was found to modulate early and late components of somatosensory and visual evoked potentials. For sensory-perceptual stages, the presence of virtual or real hands was found to amplify attention-related negativity on the somatosensory N140 and cross-modal interaction in somatosensory and visual P200. For postperceptual stages, an amplified N200 component was obtained in somatosensory and visual evoked potentials, indicating increased response inhibition in response to nontarget stimuli. The effect of somatosensory, but not visual, N200 enhanced when the virtual hands were present. The findings suggest that bodily presence affects sustained cross-modal spatial attention between vision and touch and that this effect is specifically present in ERPs related to early- and late-sensory processing, as well as response inhibition, but do not affect later attention and memory-related P3 activity. Finally, the experiments provide commeasurable scenarios for the estimation of the signal and noise ratio to quantify effects related to the use of a head mounted display (HMD). However, despite valid a-priori reasons for fearing signal interference due to a HMD, we observed no significant drop in the robustness of our ERP measurements.

A kernel-learning approach to semi-supervised clustering with relative distance comparisons

Abstract: We consider the problem of clustering a given dataset into k clusters subject to an additional set of constraints on relative distance comparisons between the data items The additional constraints are meant to reflect side-information that is not expressed in the feature vectors, directly Relative comparisons can express structures at finer level of detail than must-link (ML) and cannot-link (CL) constraints that are commonly used for semi-supervised clustering Relative comparisons are particularly useful in settings where giving an ML or a CL constraint is difficult because the granularity of the true clustering is unknown Our main contribution is an efficient algorithm for learning a kernel matrix using the log determinant divergence (a variant of the Bregman divergence) subject to a set of relative distance constraints Given the learned kernel matrix, a clustering can be obtained by any suitable algorithm, such as kernel k-means We show empirically that kernels found by our algorithm yield clusterings of higher quality than existing approaches that either use ML/CL constraints or a different means to implement the supervision using relative comparisons