The Flip-Flop Neuron – A Memory Efficient Alternative for Solving Challenging Sequence Processing and Decision Making Problems

Sequential decision making tasks that require information integration over extended durations of time are challenging for several reasons including the problem of vanishing gradients, long training times and significant memory requirements. To this end we propose a neuron model fashioned after the JK flip-flops in digital systems. A flip-flop is a sequential device that can store state information of the previous history. We incorporate the JK flip-flop neuron into several deep network architectures and apply the networks to difficult sequence processing problems. The proposed architectures include flip-flop neural networks (FFNNs), bidirectional flip-flop neural networks (BiFFNNs), convolutional flip-flop neural networks (ConvFFNNs), and bidirectional convolutional flip-flop neural networks (BiConvFFNNs). Learning rules of proposed architectures have also been derived. We have considered the most popular benchmark sequential tasks like signal generation, sentiment analysis, handwriting generation, text generation, video frame prediction, lung volume prediction, and action recognition to evaluate the proposed networks. Finally, we compare the results of our networks with the results from analogous networks with Long Short-Term Memory (LSTM) neurons on the same sequential tasks. Our results show that the JK flip-flop networks outperform the LSTM networks significantly or marginally on all the tasks, with only half of the trainable parameters.

The Flip-flop neuron – A memory efficient alternative for solving challenging sequence processing and decision making problems

Sequential decision making tasks that require information integration over extended durations of time are challenging for several reasons including the problem of vanishing gradients, long training times and significant memory requirements. To this end we propose a neuron model fashioned after the JK flip-flops in digital systems. A flip-flop is a sequential device that can store state information of the previous history. We incorporate the JK flip-flop neuron into several deep network architectures and apply the networks to difficult sequence processing problems. The proposed architectures include flip-flop neural networks (FFNNs), bidirectional flip-flop neural networks (BiFFNNs), convolutional flip-flop neural networks (ConvFFNNs), and bidirectional convolutional flip-flop neural networks (BiConvFFNNs). Learning rules of proposed architectures have also been derived. We have considered the most popular benchmark sequential tasks like signal generation, sentiment analysis, handwriting generation, text generation, video frame prediction, lung volume prediction, and action recognition to evaluate the proposed networks. Finally, we compare the results of our networks with the results from analogous networks with Long Short-Term Memory (LSTM) neurons on the same sequential tasks. Our results show that the JK flip-flop networks outperform the LSTM networks significantly or marginally on all the tasks, with only half of the trainable parameters.

Improving Coarse Grain Models of Protein Folding through Weighting of Polar-polar/hydrophobic-hydrophobic Interactions into Crowded Spaces

In this piece of work were tested 7 Hydrophobic-Polar sequences in two types of 2D-square space lattices, homogeneous and correlated, the latter simulating molecular crowding included as a geometric boundary restriction. The optimization of the 2D structures was carried out using a variant of Dill's model, inspired by the convex function, which takes into account both the hydrophobic (Dill’s model) and polar interactions, aimed to include more structural information to reach better folding solutions. While using correlated networks, the degrees of freedom in the folding of sequences were limited, and as a result in all cases more successful structural trials were found in comparison to the homogeneous lattice. In particular, the S5 sequence turned out to be the most difficult sequence of the seven folded, this perhaps due to the intrinsic i) degrees of freedom and ii) motifs of the expected 2D HP structure. Regarding S2 and S6 sequences, although optimal folding was not achieved for neither of the two approaches, folding with correlated network approach not only produced better results than homogeneous space, but for both sequences the best values found with crowding were very close to the expected optimal fitness. The sequences S1-S4 and S6 were better folded with medium lattice units for the correlated media, instead, S5 and S7 were better folded with a bit larger degree of lattice unit, revealing that depending on the degrees of freedom and particular folding motifs in each sequence would require particular crowding to achieve better folding. Finally, we claim that in all folded sequences in crowded spaces achieve better results than homogeneous ones.

Novel Purification Process for Amyloid Beta Peptide(1-40)

Amyloid beta peptide (Aβ)-related studies require an adequate supply of purified Aβ peptide. However, Aβ peptides are “difficult sequences” to synthesize chemically, and low yields are common due to aggregation during purification. Here, we demonstrate an easier synthesis, deprotection, reduction, cleavage, and purification process for Aβ(1-40) using standard 9-fluorenylmethyloxycarbonyl (Fmoc)-protected amino acids and solid-phase peptide synthesis (SPPS) resin [HMBA (4-hydroxymethyl benzamide) resin] that provides higher yields of Aβ(1-40) than previous standard protocols. Furthermore, purification requires a similar amount of time as conventional purification processes, although the peptide must be cleaved from the resin immediately prior to purification. The method described herein is not limited to the production of Aβ(1-40), and can be used to synthesize other easily-oxidized and aggregating sequences. Our proposed methodology will contribute to various fields using “difficult sequence” peptides, such as pharmaceutical and materials science, as well as research for the diagnosis and treatment of protein/peptide misfolding diseases.

The Power of Monsanto's Stone

There is no point in rushing because, in fact, people are going nowhere in particular. No matter how arduous the observation, in a slow and difficult sequence, people are always in the same place: in the countryside. There, people were yesterday and will be tomorrow. The landscape is a singularity with no limits: each tree, each granite boulder, each course presents infinite perspectives indistinguishable from each other. Sustainable landscape will be a mandatory topic in the twenty-first century and will influence the interventions in open spaces. These new paradigms will allow a healthier environment, where the relation of architecture and environmental comfort is present. In this sense, the chapter addresses aspects of the environment in its relationship with living culture; studies construction techniques with a lower environmental impact; and develops adaptive strategies of “sustainable project” for ordering and appropriation of habitable space according to the principles of cultural, economic, and environmental preservation.

The Power of Monsanto's Stone

There is no point in rushing because, in fact, people are going nowhere in particular. No matter how arduous the observation, in a slow and difficult sequence, people are always in the same place: in the countryside. There, people were yesterday and will be tomorrow. The landscape is a singularity with no limits: each tree, each granite boulder, each course presents infinite perspectives indistinguishable from each other. Sustainable landscape will be a mandatory topic in the twenty-first century and will influence the interventions in open spaces. These new paradigms will allow a healthier environment, where the relation of architecture and environmental comfort is present. In this sense, the chapter addresses aspects of the environment in its relationship with living culture; studies construction techniques with a lower environmental impact; and develops adaptative strategies of “sustainable project” for ordering and appropriation of habitable space according to the principles of cultural, economic, and environmental preservation.

A complex obstetric case

A 33-year-old primigravida presented with severe sepsis, severe pre-eclampsia, peripartum cardiomyopathy and the haemolysis, elevated liver enzymes and low platelets syndrome manifesting over the course of 24 h causing a diagnostic conundrum and a difficult sequence of physiological problems to overcome. We describe a previously unreported sequence of events involving a pre-eclamptic, septic parturient to improve anaesthetic and intensive care physician awareness of confounding factors that complicate assessment and management of these patients.