State-of-the-art sequence labeling systems traditionally require large amounts of task-specific knowledge in the form of handcrafted features and data pre-processing. In this paper, we introduce a novel neutral network architecture that benefits from both word-and character-level representations automatically, by using combination of bidirectional LSTM, CNN and CRF. Our system is truly end-to-end, requiring no feature engineering or data pre-processing, thus making it applicable to a wide range of sequence labeling tasks. We evaluate our system on two data sets for two sequence labeling tasks -- Penn Treebank WSJ corpus for part-of-speech (POS) tagging and CoNLL 2003 corpus for named entity recognition (NER). We obtain state-of-the-art performance on both datasets -- 97.55% accuracy for POS tagging and 91.21% F1 for NER. 1 Introduction Linguistic sequence labeling, such as part-of- speech (POS) tagging and named entity recognition (NER), is one of the first stages in deep language understanding and its importance has been well recognized in the natural language processing community. Most traditional high performance sequence labeling models are linear statistical models, including Hidden Markov Models (HMM) and Conditional Random Fields (CRF) (Ratinov and Roth, 2009; Passos et al., 2014; Luo et al., 2015), which rely heavily on handcrafted features and task-specific resources. For example, English POS taggers benefit from carefully designed word spelling features; orthographic features and external resources such as gazetteers are widely used in NER. However, such task-specific knowledge is costly to develop (Ma and Xia, 2014), making sequence labeling models difficult to adapt to new tasks or new domains. In the past few years, nonlinear neural networks with as input distributed word representations, also known as word embeddings, have been broadly applied to NLP problems with great success.

Many methods have been proposed for detecting emerging events in text streams using topic modeling. However, these methods have shortcomings that make them unsuitable for rapid detection of locally emerging events on massive text streams. We describe Spatially Compact Semantic Scan (SCSS) that has been developed specifically to overcome the shortcomings of current methods in detecting new spatially compact events in text streams. SCSS employs alternating optimization between using semantic scan (Liu and Neill (2011)) to estimate contrastive foreground topics in documents, and discovering spatial neighborhoods (Shao et al. (2011)) with high occurrence of documents containing the foreground topics. We evaluate our method on Emergency Department chief complaints dataset (ED dataset) to verify the effectiveness of our method in detecting real-world disease outbreaks from free-text ED chief complaint data.

Variational inference (VI) combined with data subsampling enables approximate posterior inference over large data sets, but suffers from poor local optima. We first formulate a deterministic annealing approach for the generic class of conditionally conjugate exponential family models. This approach uses a decreasing temperature parameter which deterministically deforms the objective during the course of the optimization. A well-known drawback to this annealing approach is the choice of the cooling schedule. We therefore introduce variational tempering, a variational algorithm that introduces a temperature latent variable to the model. In contrast to related work in the Markov chain Monte Carlo literature, this algorithm results in adaptive annealing schedules. Lastly, we develop local variational tempering, which assigns a latent temperature to each data point; this allows for dynamic annealing that varies across data. Compared to the traditional VI, all proposed approaches find improved predictive likelihoods on held-out data.

The recovery of the rocket module proved once again that delivering payloads into deep orbit could be much cheaper in the near future. But this idea is still in its experimental stage and would require repeated successes to become part of normal operating procedures in 21st century space transport. SpaceX plans to use one of its four recovered first-stage rockets in a mission later this year. The rocket recovery was part of a successful mission to deliver an Asian communications satellite into so-called supersynchronous orbit, a position that puts a satellite more than 22,000 miles above the earth's surface in a way where it synchronizes with the planet's orbit in order to remain above the same area at all times. The satellite, Thaicom 8, will service communications and data transfer needs in Thailand, India and East Africa, according to nasaspaceflight.com.

Harlan also noted in the comment below that any truncated kernel density estimator (KDE) from density() in R does not integrate to 1 over its support set. Thanks to Julian Richer Daily for suggesting on AnalyticBridge to scale any truncated kernel density estimator (KDE) from density() by its integral to get a KDE that integrates to 1 over its support set. I have used my own function for trapezoidal integration to do so, and this has been added below. I thank everyone for your patience while I took the time to write a post about numerical integration before posting this correction. I was in the process of moving between jobs and cities when Harlan first brought this issue to my attention, and I had also been planning a major expansion of this blog since then.

Can computers be truly creative? More specifically, can people bestow upon machines what we know as creativity and have the machines thinking creatively? Google knows that an answer does not come easily and some people may argue that the answer is hairy. Do all people agree on what makes creativity creativity? Depending on what kind of definition you go by, if you build software that can take a note sequence and turn it into a melody by finding patterns where do you place it on the scale of creativity?

The danger-sensing abilities of the newly developed robot system far exceed those of the Robot in the classic TV series Lost in Space. The danger-sensing abilities of the newly developed robot system far exceed those of the Robot in the classic TV series Lost in Space. Researchers are developing a system to teach robots how to feel pain. That might seem counterintuitive, as IEEE Spectrum points out. After all, "One of the most useful things about robots is that they don't feel pain."

Dream interpretation is the process of assigning meaning to dreams. In many ancient societies, such as those of Egypt and Greece, dreaming was considered a supernatural communication or a means of divine intervention, whose message could be unravelled by people with certain powers. In modern times, various schools of psychology and neurobiology have offered theories about the meaning and purpose of dreams. Most people currently appear to interpret dream content according to the Freudian theory of dreams in countries, as found by a study conducted in the United States, India, and South Korea.[1] People appear to believe dreams are particularly meaningful: they assign more meaning to dreams than to similar waking thoughts. For example, people report they would be more likely to cancel a trip they had planned that involved a plane flight if they dreamt of their plane crashing the night before than if they thought of their plane crashing the night before or the Department of Homeland Security issued a Federal warning.[1] However, people do not attribute equal importance to all dreams.

Who needs Art critics when a computer can do the job? Visitors to the Tate will be invited to access an Artificial Intelligence (AI) programme which uses algorithms to explain the relevance of works in the collection. "We can't wait to begin working with Tate, Microsoft and a talented team of AI specialists to create this living, seeing, algorithm." Tate Britain has awarded the 15,000 IK prize and a 90,000 production budget to the Italian team behind Recognition, a research project which will merge AI and art, to "uncover the hidden links between current events and art from the Tate collection." Supported by Microsoft, the Fabrica team, based in Treviso, will use powerful algorithms and "machine learning" to search through Tate's vast digital collection and archive and news images of current events, unearthing "hidden relationships between how the world has been represented in image form, in the past and present."

A Tesla Model S driver has crashed into the back of a van while the car's autopilot, active cruise control and automatic emergency brake were activated. The driver, from Zurich, has blamed the crash on Tesla and claims the entire front of his car must be replaced. Many have been quick to point out that the autopilot system is in a testing phase, and Tesla warns all drivers that they should not surrender full control to the car. Since the video has gained popularity online, the driver - who wishes not the be named - has said the problem was not with autopilot, but with the automatic emergency brake and cruise control. A Tesla Model S driver has crashed into the back of a van while the car's autopilot and automatic emergency brake were activated.