Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

Link prediction in complex networks has attracted increasing attention from both physical and computer science communities. The algorithms can be used to extract missing information, identify spurious interactions, evaluate network evolving mechanisms, and so on. This article summaries recent progress about link prediction algorithms, emphasizing on the contributions from physical perspectives and approaches, such as the random-walk-based methods and the maximum likelihood methods. We also introduce three typical applications: reconstruction of networks, evaluation of network evolving mechanism and classification of partially labeled networks. Finally, we introduce some applications and outline future challenges of link prediction algorithms.

/pdf/link-prediction-in-complex-networks-a-survey-fcztp3ymq3.pdf

Link prediction in complex networks: A survey

인공고관절의 세라믹 볼 헤드의 기계적 안정성 평가를 위한 3 차원 유한요소 해석

https://people.mpi-inf.mpg.de/~kberberi/publications/2010-mpii-tra.pdf

YAGO2: A spatially and temporally enhanced knowledge base from Wikipedia

https://pure.mpg.de/pubman/item/item_1819068_3/component/file_1840695/MPI-I-2007-5-003.pdf

YAGO: A Large Ontology from Wikipedia and WordNet

Human beings always engage in a vast range of activities and tasks that demonstrate their ability to adapt to different scenarios. Any human activity can be represented as a temporal sequence of actions performed to achieve a certain goal. Unlike the time series datasets extracted from electronics or machines, these action sequences are highly disparate in their nature -- the time to finish a sequence of actions can vary between different persons. Therefore, understanding the dynamics of these sequences is essential for many downstream tasks such as activity length prediction, goal prediction, next action recommendation, etc. Existing neural network-based approaches that learn a continuous-time activity sequence (or CTAS) are limited to the presence of only visual data or are designed specifically for a particular task, i.e., limited to next action or goal prediction. In this paper, we present ProActive, a neural marked temporal point process (MTPP) framework for modeling the continuous-time distribution of actions in an activity sequence while simultaneously addressing three high-impact problems -- next action prediction, sequence-goal prediction, and end-to-end sequence generation. Specifically, we utilize a self-attention module with temporal normalizing flows to model the influence and the inter-arrival times between actions in a sequence. In addition, we propose a novel addition over the ProActive model that can handle variations in the order of actions, i.e., different methods of achieving a given goal. We demonstrate that this variant can learn the order in which the person or actor prefers to do their actions. Extensive experiments on sequences derived from three activity recognition datasets show the significant accuracy boost of ProActive over the state-of-the-art in terms of action and goal prediction, and the first-ever application of end-to-end action sequence generation.

Tapestry of Time and Actions: Modeling Human Activity Sequences using Temporal Point Process Flows

Two common types of tasks on Knowledge Bases have been studied -- single link prediction (Knowledge Base Completion) and path query answering. However, our analysis of user queries on a real-world knowledge base reveals that a significant fraction of queries specify paths using regular expressions(regex). Such regex queries cannot be handled by any of the existing link prediction or path query answering models. In response, we present Regex Query Answering, the novel task of answering regex queries on incomplete KBs. We contribute two datasets for the task, including one where test queries are harvested from actual user querylogs. We train baseline neural models for our new task and propose novel ways to handle disjunction and Kleene plus regex operators.

Knowledge Base Inference for Regular Expression Queries.

We introduce TechTrack, a new dataset for tracking entities in technical procedures. The dataset, prepared by annotating open domain articles from WikiHow, consists of 1351 procedures, e.g., "How to connect a printer", identifies more than 1200 unique entities with an average of 4.7 entities per procedure. We evaluate the performance of state-of-the-art models on the entity-tracking task and find that they are well below the human annotation performance. We describe how TechTrack can be used to take forward the research on understanding procedures from temporal texts.

Tracking entities in technical procedures - a new dataset and baselines.

In applications arising in massive on-line social networks, biological networks, and knowledge graphs it is often required to find shortest length path between two given nodes. Recent results have addressed the problem of computing either exact or good approximate shortest path distances efficiently. Some of these techniques also return the path corresponding to the estimated shortest path distance fast. Many of the real-world graphs are edge-labeled graphs, i.e., each edge has a label that denotes the relationship between the two vertices connected by the edge. However, none of the techniques for estimating shortest paths work very well when we have additional constraints on the labels associated with edges that constitute the path. In this work, we define the problem of retrieving shortest length path between two given nodes which also satisfies user-provided constraints on the set of edge labels involved in the path. We have developed SkIt index structure, which supports a wide range of label constraints on paths, and returns an accurate estimation of the shortest path that satisfies the constraints. We have conducted experiments over graphs such as social networks, and knowledge graphs that contain millions of nodes/edges, and show that SkIt index is fast, accurate in the estimated distance and has a high recall for paths that satisfy the constraints.

Label constrained shortest path estimation on large graphs

Evolving content of the Web is being accumulated over time into Web archival collections. This creates the need for time travel search to explore the dynamics of the content. Text analytics has also a key role in exploring interesting information in text collections. Moreover, frequent phrase mining, a special case of text analytics, is an important analytical task that is motivated by the need of knowledge on frequent phrases in various areas of computer science, such as information retrieval and machine translation etc. However, time travel search and frequent phrase mining have to be conducted on increasingly large-scale data. Distributed approaches such as MapReduce, which is mainly designed to work on vast amount of text, can be utilized in this case. We address two separate problems in this thesis. The first problem is that time travel inverted index, which enables searching on the time dimension, is proposed in centralized setting. In our work, we parallelize the construction of time travel inverted index using MapReduce, having a distributed index as an end product. The second problem is that finding frequent phrases through näıve counting, even in MapReduce, is a time consuming task, because data to be processed gets much larger in size, when phrases are considered. As our work, we present partitioned approximate phrase counting, a very fast way to retrieve most of the frequent phrases together with their counts out of a collection to enable interactive analysis of the content. Included in this, we propose and a novel technique, partitioned in-mapper combining, which enables us to aggregate data in memory correctly, even though the data to be aggregated is larger than the available memory. Evaluation of experiments on New York Times Annotated Corpus, which contains roughly 2 million documents, show that our approach works at least 2 times faster as compared to näıve approach. It finds more than 90% of frequent phrases with high precision. Moreover, it is able to find all highly frequent phrases exactly, along with their accurate counts. Furthermore, by a quick second pass on the data, we precisely provide most of the frequent phrases with their corresponding true counts, still being significantly faster than näıve approach. Figure 1: Tag cloud of phrases in New York Times Annotated Corpus (1987 2007)

Srikanta Bedathur

Papers

Tapestry of Time and Actions: Modeling Human Activity Sequences using Temporal Point Process Flows

Knowledge Base Inference for Regular Expression Queries.

Tracking entities in technical procedures - a new dataset and baselines.

Label constrained shortest path estimation on large graphs

Distributed Analytics over Web Archives