Avoiding Catastrophe: Active Dendrites Enable Multi-Task Learning in Dynamic Environments.

A key challenge for AI is to build embodied systems that operate in dynamically changing environments. Such systems must adapt to changing task contexts and learn continuously. Although standard deep learning systems achieve state of the art results on static benchmarks, they often struggle in dynamic scenarios. In these settings, error signals from multiple contexts can interfere with one another, ultimately leading to a phenomenon known as catastrophic forgetting. In this article we investigate biologically inspired architectures as solutions to these problems. Specifically, we show that the biophysical properties of dendrites and local inhibitory systems enable networks to dynamically restrict and route information in a context-specific manner. Our key contributions are as follows: first, we propose a novel artificial neural network architecture that incorporates active dendrites and sparse representations into the standard deep learning framework. Next, we study the performance of this architecture on two separate benchmarks requiring task-based adaptation: Meta-World, a multi-task reinforcement learning environment where a robotic agent must learn to solve a variety of manipulation tasks simultaneously; and a continual learning benchmark in which the model's prediction task changes throughout training. Analysis on both benchmarks demonstrates the emergence of overlapping but distinct and sparse subnetworks, allowing the system to fluidly learn multiple tasks with minimal forgetting. Our neural implementation marks the first time a single architecture has achieved competitive results in both multi-task and continual learning settings. Our research sheds light on how biological properties of neurons can inform deep learning systems to address dynamic scenarios that are typically impossible for traditional ANNs to solve.

Study of morphological properties and rheological parameters of cellulose nanofibrils of cocoa shell (Theobroma cacao L.).

Cocoa shell was evaluated as a precursor for cellulose nanofibrils (NFCs) using mechanical defibrillation. Its morphology was analysed using optical microscopy and scanning electron microscopy with field emission. Rheological and mechanical behaviour were evaluated through flow curves with a strain rate ranging from 0 to 300 s-1 at 25 °C and by means of oscillatory frequency sweeps (0.01 Hz-10 Hz) and shear stress (3 Pa). The thermal-mechanical behaviour was determined by a temperature sweep with a heating rate of 3 °C min-1 and a temperature range of 25 °C-100 °C. Micrographs identified the presence of protoxilem with a mean diameter of 23.34 nm. The flow curve showed the characteristic behaviour of a pseudoplastic fluid. The storage module (G') and the loss modulus (G″) were dependent on the frequency applied, indicating that the material exhibits a weak gel characteristic. The viscoelastic characteristics were influenced by temperature. Therefore, cocoa shell is a new alternative in the production of nanocellulose.

Production of xylitol and bio-detoxification of cocoa pod husk hemicellulose hydrolysate by Candida boidinii XM02G.

The use of cocoa pod husk hemicellulose hydrolysate (CPHHH) was evaluated for the production of xylitol by Candida boidinii XM02G yeast isolated from soil of cocoa-growing areas and decaying bark, as an alternative means of reusing this type of waste. Xylitol was obtained in concentrations of 11.34 g.L-1, corresponding to a yield (Yp/s) of 0.52 g.g-1 with a fermentation efficiency (ε) of 56.6%. The yeast was tolerant to inhibitor compounds present in CPHHH without detoxification in different concentration factors, and was able to tolerate phenolic compounds at approximately 6 g.L-1. The yeast was also able to metabolize more than 99% (p/v) of furfural and hydroxymethylfurfural present in the non-detoxified CPHHH without extension of the cell-growth lag phase, showing the potential of this microorganism for the production of xylitol. The fermentation of cocoa pod husk hydrolysates appears to provide an alternative use which may reduce the impact generated by incorrect disposal of this waste.

Enzymatic saccharification of lignocellulosic residues using cellulolytic enzyme extract produced by Penicillium roqueforti ATCC 10110 cultivated on residue of yellow mombin fruit.

The aim of this work was to enzymatic saccharification of food waste was performed by crude enzymatic cellulolytic extract produced by P. roqueforti cultivated in yellow mombin residue. The best yield of reducing sugars (259.45mgg-1) was achieved with sugarcane bagasse after 4h; the hydrolysis of corn cob, rice husk and peanut hull resulted in yields around 128-180mgg-1. The addition of 10mmolL-1 of Mn2+ potentiated the saccharification of sugarcane bagasse, in about 86%. The temperature and substrate (sugarcane bagasse) concentration parameters were optimized using a Doehlert Design and, a maximum sugar yield of 662.34±26.72mgg-1 was achieved at 62.40°C, 0.22% (w/v) of substrate, with the addition of Mn2+. Sugar yield was significantly high when compared to previous studies available in scientific literature, suggesting the use of crude cellulolytic supplemented with Mn2+ an alternative and promising process for saccharification of sugarcane bagasse.