Solving Sequential Text Classification as Board-Game Playing

Sequential Text Classification (STC) aims to classify a sequence of text fragments (e.g., words in a sentence or sentences in a document) into a sequence of labels. In addition to the intra-fragment text contents, considering the inter-fragment context dependencies is also important for STC. Previous sequence labeling approaches largely generate a sequence of labels in left-to-right reading order. However, the need for context information in making decisions varies across different fragments and is not strictly organized in a left-to-right order. Therefore, it is appealing to label the fragments that need less consideration of context information first before labeling the fragments that need more. In this paper, we propose a novel model that labels a sequence of fragments in jumping order. Specifically, we devise a dedicated board-game to develop a correspondence between solving STC and board-game playing. By defining proper game rules and devising a game state evaluator in which context clues are injected, at each round, each player is effectively pushed to find the optimal move without position restrictions via considering the current game state, which corresponds to producing a label for an unlabeled fragment jumpily with the consideration of the contexts clues. The final game-end state is viewed as the optimal label sequence. Extensive results on three representative datasets show that the proposed approach outperforms the state-of-the-art methods with statistical significance.

Mechanisms of Human intelligence — From RNA and Synapse to Broadband

Simple thought has been explained by the action-potential (AP) system with its synapses. In contrast, in-depth details for “Declarative” intellectual thought have been a complete mystery because (it is argued here) its main underlying mechanism is fundamentally different. Declarative thinking depends heavily on linear coding based on digit-like elements — something which an unaided AP system could never offer......Looking instead to psychology, Piaget (1920s) proposed basic units of action-sequences (“schèmes” whereby one could mentally construct object-concepts). There is now evidence that some ncRNA serves this verb-like action-coding role. — (Other ncRNA demonstrably serves as adjectival/adverbial “regulators” — while the remaining ≈3% of RNA encodes physical structures, the traditional noun-like role). If valid, then:–•NEW FOCUS ONTO ULTRAMICRO: — The whole Piagetian structure-coding for a concept could fit into one of the many 125nm capsids (“granules”). Moreover, many more concepts (and duplicates) could fit into a cell-body. — The vast abundance of coding-sites would allow comprehensive “wasteful” rapid use of Jerneian/Darwinian selection instead of problematic “writing down” of new learnings. — Estimates of memory-capacity increase vastly. — And hereditary-schèmes obviously explain inherited behaviour-traits. — Piaget’s other theory about develop¬mental stages also seems compatible.•Quantum-constraints ensure that such micro-sites would USE OPTICAL FREQUENCY signalling. That opens the way to greatly enhanced “Gigabit” rates, and optical-interference tricks.•MYELIN gets the EXTRA ROLE OF OPTIC-CABLE.So nerve-fibres become seen as simultan¬eous paths for two different types of signal (also demonstrated by Sun-et-al, 2010), with AP still dominant in some roles, but subservient to “UPE” optical signals elsewhere.•LOCATING MEMORIES? Choosing the right address means selecting some sort of “phone-number or numbered plug-socket.” That is best provided as an “address-label” sequence on the transmitted version of the schème-coding — in which case, actual destination-location may be less important.•“Moving-house” TO CORTICAL LONG-TERM MEMORY. The memory-move must preserve existing (i) memory-structure, and (ii) links to distant static archives. This “impossible” task would seem feasible if memories are actually held within individual cells (as above). There is indeed lifelong flow of such neurons in some mammals; but these flows seem to cease in adulthood for humans and dolphins! So the search continues.

Indexing labeled sequences

Background Labels are a way to add some information on a text, such as functional annotations such as genes on a DNA sequences. V(D)J recombinations are DNA recombinations involving two or three short genes in lymphocytes. Sequencing this short region (500 bp or less) produces labeled sequences and brings insight in the lymphocyte repertoire for onco-hematology or immunology studies. Methods We present two indexes for a text with non-overlapping labels. They store the text in a Burrows–Wheeler transform (BWT) and a compressed label sequence in a Wavelet Tree. The label sequence is taken in the order of the text (TL-index) or in the order of the BWT (TLBW-index). Both indexes need a space related to the entropy of the labeled text. Results These indexes allow efficient text–label queries to count and find labeled patterns. The TLBW-index has an overhead on simple label queries but is very efficient on combined pattern–label queries. We implemented the indexes in C++ and compared them against a baseline solution on pseudo-random as well as on V(D)J labeled texts. Discussion New indexes such as the ones we proposed improve the way we index and query labeled texts as, for instance, lymphocyte repertoire for hematological and immunological studies.

Subcutaneous hydrocortisone administration for emergency use in adrenal insufficiency

ObjectiveEvaluation of the pharmacokinetics and safety of s.c. hydrocortisone injection for use in adrenal emergency.DesignSingle-center, open-label, sequence-randomized, crossover study in a tertiary care center.Patients and methodsTwelve patients with chronic Addison's disease. Comparison of hydrocortisone pharmacokinetics after s.c. and i.m. injection (100 mg) and after s.c. administration of sodium chloride (0.9%) respectively at three different visits.Main outcome measure: maximum serum cortisol (Cmax), time to Cmax (tmax), and time to serum cortisol >36 μg/dl (tserum cortisol >36 μg/dl) after s.c. administration compared with i.m. administration, safety, and patient preference.ResultsSerum cortisol increased rapidly and substantially after both i.m. and s.c. injections (Cmax: 110±29 vs 97±28 μg/dl, P=0.27, tmax: 66±51 vs 91±34 min, P=0.17, and tserum cortisol >36 μg/dl: 11±5 vs 22±11 min, P=0.004 respectively). Both i.m. and s.c. injections were well tolerated. Eleven (91.7%) patients preferred s.c. injection, whereas one patient did not have any preference.ConclusionsS.c. administration of 100 mg hydrocortisone shows excellent pharmacokinetics for emergency use with only a short delay in cortisol increase compared with i.m. injection. It has a good safety profile and is preferred by patients over i.m. injection.

The Parenthetical Function in A la recherche du temps perdu

This essay examines parentheses on the level of individual narrative sequences in A la recherche du temps perdu. A narrative sequence is a limited, linear series of events that can be subsumed under a single label. “Sequence-level parenthesis” may be defined as an independent textual segment inserted between two contiguous moments in the sequence, interrupting its forward movement. The unique feature of Proust's sequence-level parentheses is the multiplicity of functions they fulfill: narrative, interpretive, associative. (1) Parentheses fill in gaps in the story, announce, prepare, or generate events to come, remind us of past events, etc. (2) They qualify or explain specific facts or situations; they also formulate generalizations or laws. (3) Parenthetical associations play a role analogous to that of the metaphor and the phenomenon of involuntary memory: they establish connections (rapprochements) between widely separated textual fragments. To borrow a term used by Proust, Spitzer, and more recently Deleuze, parentheses are the transversales of textual multiplicity.