Academic literature on the topic 'Feed forward layer'
Create a spot-on reference in APA, MLA, Chicago, Harvard, and other styles
Consult the lists of relevant articles, books, theses, conference reports, and other scholarly sources on the topic 'Feed forward layer.'
Next to every source in the list of references, there is an 'Add to bibliography' button. Press on it, and we will generate automatically the bibliographic reference to the chosen work in the citation style you need: APA, MLA, Harvard, Chicago, Vancouver, etc.
You can also download the full text of the academic publication as pdf and read online its abstract whenever available in the metadata.
Journal articles on the topic "Feed forward layer"
1
Safi, Youssef, and Abdelaziz Bouroumi. "Evolutionary single hidden-layer feed forward networks." International Journal of Innovative Computing and Applications 6, no. 2 (2014): 73. http://dx.doi.org/10.1504/ijica.2014.066497.
Full text
2
Hasbi, Yasin, Warsito Budi, and Santoso Rukun. "Feed Forward Neural Network Modeling for Rainfall Prediction." E3S Web of Conferences 73 (2018): 05017. http://dx.doi.org/10.1051/e3sconf/20187305017.
Full textAbstract:
Prediction of rainfall data by using Feed Forward Neural Network (FFNN) model is proposed. FFNN is a class of neural network which has three layers for processing. In time series prediction, including in case of rainfall data, the input layer is the past values of the same series up to certain lag and the output layer is the current value. Beside a few lagged times, the seasonal pattern also considered as an important aspect of choosing the potential input. The autocorrelation function and partial autocorrelation function patterns are used as aid of selecting the input. In the second layer called hidden layer, the logistic sigmoid is used as activation function because of the monotonic and differentiable. Processing is done by the weighted summing of the input variables and transfer process in the hidden layer. Backpropagation algorithm is applied in the training process. Some gradient based optimization methods are used to obtain the connection weights of FFNN model. The prediction is the output resulting of the process in the last layer. In each optimization method, the looping process is performed several times in order to get the most suitable result in various composition of separating data. The best one is chosen by the least mean square error (MSE) criteria. The least of in-sample and out-sample predictions from the repeating results been the base of choosing the best optimization method. In this study, the model is applied in the ten-daily rainfall data of ZOM 136 Cokrotulung Klaten. Simulation results give a consecution that the more complex architecture is not guarantee the better prediction.
3
Svozil, Daniel, Vladimír Kvasnicka, and Jir̂í Pospichal. "Introduction to multi-layer feed-forward neural networks." Chemometrics and Intelligent Laboratory Systems 39, no. 1 (1997): 43–62. http://dx.doi.org/10.1016/s0169-7439(97)00061-0.
Full text
4
Behunkou, U. I. "Loan classification using a feed-forward neural network." Informatics 21, no. 1 (2024): 83–104. http://dx.doi.org/10.37661/1816-0301-2024-21-1-83-104.
Full textAbstract:
Objectives. The purpose of the study is to construct and study the use of a feed-forward neural network to solve the problem of loan classification, as well as to conduct a comparative analysis of the neural networkbased approach with the existing approach based on logistic regression.Methods. Based on a feed-forward neural network using historical data on loans issued, the following metrics are calculated: cost function, Accuracy, Precision, Recall, and measure, calculated on Precision and Recall values. Polynomial parameters and the principal component method are used to determine the optimal set of input data for the studied neural network.Results. The impact of data normalization on the final result was analyzed, the influence of the number of units in the hidden layer on the outcome was evaluated using a two-stage method and the Monte Carlo method, the effect of balanced data use was determined, the optimal threshold value for output layer of the neural network under investigation was calculated, the optimal activation function for the hidden layer units was found, the effect of increasing input indicators through missing values imputation and the use of polynomials of varying degrees was studied and the redundancy in the existing set of input indicators was analyzed.Conclusion. Based on the results of the research, we can conclude that the use of a direct distribution network to solve problems of loan classification is appropriate. Compared to logistic regression, implementing a solution using a feed-forward neural network requires more time and computing resources. However, the obtained most important values of Accuracy and measure are higher than those calculated using logistic regression [1].
5
Hayat, Safdar. "Phone numbers classificationwith feed-forward neural networks." Foundation University Journal of Engineering and Applied Sciences <br><i style="color:yellow;">(HEC Recognized Y Category , ISSN 2706-7351)</i> 1, no. 2 (2021): 1–5. http://dx.doi.org/10.33897/fujeas.v1i2.340.
Full textAbstract:
A neural network (NN)-based method for phone number classification or recognition is provided in this paper. The used network is a one-hidden-layer multilayer perceptron (MLP) classifier. Its training is based on backpropagation learning. I present the results of a Feed Forward Neural Network trained to classify phone numbers into four categories: Different training data were pre-processed and then tested to distinguish between four classes/patterns of phone numbers in order to train the FFNN. My goal is to provide a coalescence of the published research in this field and to arouse further research interest in and efforts to research the identified topics.
6
Karakida, Ryo, Yasuhiko Igarashi, Kenji Nagata, and Masato Okada. "Inter-Layer Correlation in a Feed-Forward Network with Intra-Layer Common Noise." Journal of the Physical Society of Japan 82, no. 6 (2013): 064007. http://dx.doi.org/10.7566/jpsj.82.064007.
Full text
7
Indurkhya, Nitin, and Sholom M. Weiss. "Heuristic configuration of single hidden-layer feed-forward neural networks." Applied Intelligence 2, no. 4 (1992): 325–31. http://dx.doi.org/10.1007/bf00058649.
Full text
8
Pham, Minh Tuan, Syouhei Nishihama, and Kazuharu Yoshizuka. "Removal of Chromium from Water Environment by Forward Osmosis System." MATEC Web of Conferences 333 (2021): 04007. http://dx.doi.org/10.1051/matecconf/202133304007.
Full textAbstract:
Forward osmosis (FO) technology has been applied for removal of chromium (Cr) from water environment. Comparison of the removal efficiency of Cr(VI) and Cr(III) was investigated by changing several operational conditions. The pH of feed solution plays an important role in rejection of Cr. The Cr(VI) rejection was increased with increasing pH, while Cr(III) rejection was stable. It also demonstrated that the rejection of Cr was higher when the membrane active layer faces the feed solution compared to the rejection when the membrane active layer faces the draw solution.
9
Pham, Minh Tuan, Syouhei Nishihama, and Kazuharu Yoshizuka. "Removal of Chromium from Water Environment by Forward Osmosis System." MATEC Web of Conferences 333 (2021): 04007. http://dx.doi.org/10.1051/matecconf/202133304007.
Full textAbstract:
Forward osmosis (FO) technology has been applied for removal of chromium (Cr) from water environment. Comparison of the removal efficiency of Cr(VI) and Cr(III) was investigated by changing several operational conditions. The pH of feed solution plays an important role in rejection of Cr. The Cr(VI) rejection was increased with increasing pH, while Cr(III) rejection was stable. It also demonstrated that the rejection of Cr was higher when the membrane active layer faces the feed solution compared to the rejection when the membrane active layer faces the draw solution.
10
Mahmood, M. F., and Z. Ahmad. "Application of Multi-Layer Feed Forward Neural Network (MLFNN) for the." Nucleus 54, no. 1 (2017): 10–15. https://doi.org/10.71330/thenucleus.2017.92.
Full textAbstract:
The geophysical formation evaluation plays a fundamental role in hydrocarbon exploration. Porosity is one of the main parameters that determine the amount of oil present in a rock formation. Accurate determination of porosity is a difficult problem due to failure in understanding of spatial porosity parameter distribution. Multi-layer feed forward neural network (MLFN) proved to be a powerful tool for mapping porosity across the whole field and proved to be a powerful tool for mapping complicated relationships in reservoir. In MLFN three layers are involved that is an input layer, an output layer and a variable number of hidden layers. Input for training eight external attributes are used which are P-impedance, S-impedance, density, fluid, lithology impedance, lamda-rho, mu-rho, and Vp/Vs. Five nodes are used in hidden layer and one output node for mapping total porosity of Badin gas field. In this study 3D cube of Badin field and 3 wells are used. The findings proved competence of multi-layer feed forward neural network in the porosity prediction process with an average error of 0.014 [v/v] and the correlation coefficient of 0.91 and helped in studying the lateral variations in the porosity along the reservoir. The A sands show same porosity values along both the well locations, while for B sand the porosity value decreases from Zaur-01 to Chakri-01 well while for C sand the porosity value increases from Zaur-01 to Chakri-01 well.
Dissertations / Theses on the topic "Feed forward layer"
1
Ahmed, Zulfiqar. "An hybrid architecture for multi-layer feed-forward neural networks." Thesis, National Library of Canada = Bibliothèque nationale du Canada, 1999. http://www.collectionscanada.ca/obj/s4/f2/dsk1/tape8/PQDD_0009/MQ52500.pdf.
Full text
2
Zheng, Gonghui. "Design and evaluation of a multi-output-layer perceptron." Thesis, University of Ulster, 1996. http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.338195.
Full text
3
Kanichay, Roby Thomas. "Feed-forward excitation of interneurons in the cerebellar granule cell layer." Thesis, University College London (University of London), 2008. http://discovery.ucl.ac.uk/1444212/.
Full textAbstract:
The cerebellum is involved in maintenance of posture and balance and coordination of voluntary movements. It has previously been shown that the inhibition of granule cells by Golgi cells, in the input layer of the cerebellar cortex, is important for normal cerebellar function. However, little is known about what determines the firing of spontaneously active Golgi cells and how intrinsic activity interacts with sensory input. In particular, the excitation of the interneuron by mossy fibres, which may mediate feed-forward inhibition of granule cells, has not been characterized. I have used immuno-histochemistry, patch-clamp recordings and imaging in acute cerebellar slices of rats to study feed-forward mossy fibre input onto Golgi cells and its downstream effects. I confirm that mossy fibres, Golgi cells and granule cells form a functional feed-forward inhibitory circuit. Anatomical analysis of the circuitry suggests that only spatially correlated inputs result in feed-forward inhibition. Activation of the pathway required synchronous activity in 4 out of the approximately 10 mossy fibres contacting a Golgi cell. These inputs can reset the timing of spontaneous Golgi cell firing with remarkably high temporal precision. I found that an interaction between fast EPSC kinetics, electronic compactness and pacemaker conductances allowed precise temporal signaling while integrating only 6 quanta across the dendritic tree of a Golgi cell. Golgi cell mean firing rate was only weakly modulated by mossy fibre input due to dominant pacemaker conductances. These results suggest that the properties of the feed-forward mossy fibre - Golgi cell - granule cell pathway are tuned to detect and signal coincident synaptic activity with high temporal precision. This provides a likely synaptic basis for precisely timed Golgi cell responses observed in vivo, which may signal the onset of sensory stimulation producing spatiotemporally correlated mossy fibre activity. These findings are discussed in the context of current models of granule cell layer processing.
4
Setyawati, Bina R. "Multi-layer feed forward neural networks for foreign exchange time series forecasting." Morgantown, W. Va. : [West Virginia University Libraries], 2005. https://eidr.wvu.edu/etd/documentdata.eTD?documentid=4180.
Full textAbstract:
Thesis (Ph. D.)--West Virginia University, 2005.<br>Title from document title page. Document formatted into pages; contains xii, 185 p. : ill. (some col.). Includes abstract. Includes bibliographical references (p. 140-146).
5
Golmohammadi, Davood. "A decision making model for evaluating suppliers by multi-layer feed forward neural networks." Morgantown, W. Va. : [West Virginia University Libraries], 2007. https://eidr.wvu.edu/etd/documentdata.eTD?documentid=5455.
Full textAbstract:
Thesis (Ph. D.)--West Virginia University, 2007.<br>Title from document title page. Document formatted into pages; contains xiii, 200 p. : ill. (some col.). Includes abstract. Includes bibliographical references (p. 143-151).
6
Ismail, T. "Feed-forward linearisation of a directly modulated semiconductor laser and broadband millimetre-wave wireless over fibre systems." Thesis, University College London (University of London), 2007. http://discovery.ucl.ac.uk/1444753/.
Full textAbstract:
This thesis is concerned with reduction of non-linear distortion in a directly modulated uncooled semiconductor laser using feed-forward compensation and investigating the performance of broadband millimetre-wave wireless over fibre systems. One of the key elements that determine the performance in a fibre optic link is the linearity of the optical source. Direct modulation of an uncooled semiconductor laser diode is a simple and cost effective solution. However, the distortion and noise generated by the laser limit the achievable dynamic range and performance in a system. Feed-forward linearisation is demonstrated at 5 GHz, the highest operating frequency reported, with 26 dB third order intermodulation distortion suppression and simultaneous noise reduction leading to enhanced spurious free dynamic range of 107 dB (1Hz). The effectiveness of feed-forward in a multi-channel system is investigated. Laser non-linearity can cause spectral re-growth and interchannel distortion that can completely mask the adjacent channel. A significant 11 dB interchannel distortion suppression and 10.5 dB power advantage is obtained compared to the non-linearised case. These results suggest that feed-forward linearisation arrangement can make a practical multi-channel or multi-operator wireless over fibre system. In the second part of this thesis the first experimental transmission of wireless data over fibre with remote millimetre-wave local oscillator delivery using a bi-directional semiconductor optical amplifier in a full duplex system with 2.2 km coarse wavelength division multiplexing fibre ring architecture is demonstrated. The use of bi-directional SOAs in place of unidirectional erbium doped fibre amplifier or unidirectional SOAs, together with the use of CWDM and optical distribution of the local oscillator signal allow substantial reduction in overall complexity and cost. Successful transmission of data over 12.8 km fibre is achieved with clear and well defined constellations and eye diagrams as well as 10.5% and 7.8 % error vector magnitude values for the downlink and uplink directions, respectively. The thesis also presents an implementation and performance of a millimetre-wave gigabit wireless over fibre system. CWDM devices such as uncooled laser diodes and passive components are used for the first time in a gigabit system allowing cost savings compared to dense WDM. This makes such solutions more attractive for millimetre-wave access systems. Optically modulated gigabit wireless data signals to and from the base stations are distributed at 5 GHz and up-converted using a remotely delivered LO source. Eye diagrams and bit error rate are measured to assess the system performance.
7
SARCIA', SALVATORE ALESSANDRO. "An Approach to improving parametric estimation models in the case of violation of assumptions based upon risk analysis." Doctoral thesis, Università degli Studi di Roma "Tor Vergata", 2009. http://hdl.handle.net/2108/1048.
Full textAbstract:
In this work, we show the mathematical reasons why parametric models fall short of providing correct estimates and define an approach that overcomes the causes of these shortfalls. The approach aims at improving parametric estimation models when any regression model assumption is violated for the data being analyzed. Violations can be that, the errors are x-correlated, the model is not linear, the sample is heteroscedastic, or the error probability distribution is not Gaussian. If data violates the regression assumptions and we do not deal with the consequences of these violations, we cannot improve the model and estimates will be incorrect forever. The novelty of this work is that we define and use a feed-forward multi-layer neural network for discrimination problems to calculate prediction intervals (i.e. evaluate uncertainty), make estimates, and detect improvement needs. The primary difference from traditional methodologies is that the proposed approach can deal with scope error, model error, and assumption error at the same time. The approach can be applied for prediction, inference, and model improvement over any situation and context without making specific assumptions. An important benefit of the approach is that, it can be completely automated as a stand-alone estimation methodology or used for supporting experts and organizations together with other estimation techniques (e.g., human judgment, parametric models). Unlike other methodologies, the proposed approach focuses on the model improvement by integrating the estimation activity into a wider process that we call the Estimation Improvement Process as an instantiation of the Quality Improvement Paradigm. This approach aids mature organizations in learning from their experience and improving their processes over time with respect to managing their estimation activities. To provide an exposition of the approach, we use an old NASA COCOMO data set to (1) build an evolvable neural network model and (2) show how a parametric model, e.g., a regression model, can be improved and evolved with the new project data.
8
Svensson, Björn. "Mät och reglerproblem liners." Thesis, Mittuniversitetet, Institutionen för elektronikkonstruktion, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:miun:diva-42535.
Full textAbstract:
Inpipes produktion av liners sker idag till stor del manuellt. I syfte att minska variansen på produktkvalitet, råvaror och öka processens hastighet undersöktes möjligheterna att elektroniskt mäta utbuktningarna som uppstod i processen. Därefter undersöktes processens egenskaper inför att använda mätvärdena i en passande reglerstruktur. Efter en urvalsprocess i samråd med Inpipe och iLogik ansågs två av fem mätmetoder intressanta att gå vidare med i projektet. Utifrån dem togs beslut att undersöka hur beröringsfria givare med ultraljud och optisk teknik presterade på att mäta utbuktningarnas höjd. Mätningar utfördes både i testmiljö och i fabrik med laser, ultraljudsgivare och manuellt. Resultat visade att ultraljudsgivaren hade störst precision där optiska givaren tenderade att ge ett konstant mätfel. Processens utmanande egenskaper inför en reglering av bulorna fastslogs till att främst innefatta långsamma hartsflöden, varierande produktlängd och ytvariationer. Trots utmaningar som vågor på produktens yta samt ultraljudsgivarens mätavvikelse på lutande produkt resulterade arbetet i att ultraljudsgivaren bör fungera tillfredställande till att mäta utbuktningarnas höjdförändringar och implementeras i en reglering. Utifrån processens egenskaper och inom ramen för denna studie anses framkoppling som en lämpad reglerstruktur, däremot krävs det fortsatta undersökningar för att bekräfta eller dementera det.<br>Inpipes production of liners is today substantially done manually, in order to reduce the variance of product quality, raw material and production speed, the possibility to electronically measure the swell of the product in the process was examined. The property of the industrial process was also examined to find a suitable regulation structure. A presentation of different methods of measurement was conducted, Inpipe chose two of them as specially interesting. It included measurement of ultrasound and optical laser sensors and decision to examine their precision on the liners swell was made. The measurement was done in controlled environment on a product dummy and in factory on the real liner. The ultrasound sensor proved to give the highest precision while the laser sensor indicated to measure a constant offset. The most challenging properties of the industrial process was determined to be slow resin flows, varying product length and surface variation on the product. Despite challenges as wave on the liners surface and the ultrasounds sensors inability to measure the liners slope, the project led to recommend the ultrasound sensor for measure the height of the liners swell and use the value in a regulation of the process. In theory feed forward regulation seems to suit the properties of the industrial process, but further work must be done to really confirm or deny that claim.
9
Sahni, Mohamed Omar. "Contribution à l'étude de techniques pour l'affinement spectral de lasers : application aux diodes à blocage de modes destinées aux télécommunications optiques cohérentes." Thesis, Rennes 1, 2018. http://www.theses.fr/2018REN1S121/document.
Full textAbstract:
Les peignes de fréquences optiques, issus de diodes à blocage de modes, font partie des candidats potentiels pour les réseaux de transmission à multiplexage en longueurs d’onde (WDM). Cependant, les modes composant leur peigne, exhibent généralement des largeurs de raie optiques relativement élevées ( 1-100 MHz), rendant ainsi incompatible leur utilisation sur un réseau WDM employant des formats de modulation avancés d’ordre supérieur. Cette thèse étudie, une solution pour palier à cette limitation. La technique utilisée, dite d’asservissement à correction aval hétérodyne, effectue un traitement du flux lumineux en sortie du laser sans agir sur ce dernier, permettant de réduire le bruit de fréquence présent sur chacune des raies et par conséquent leur largeur de raie optique. Dans une première approche, la technique est appliquée à un laser mono-fréquence. Cela a permis d’une part de valider son fonctionnement et d’autre part d’identifier les limites intrinsèques du dispositif expérimental mis en place. Ainsi, nous démontrons que le niveau de bruit de fréquence minimum permis par notre système, correspond à un spectre optique de largeur de raie optique instantanée de 50 Hz et une largeur de raie de 1,6 kHz pour un temps d’observation de 10 ms. La technique est par la suite appliquée à une diode à blocage de modes actif. Le peigne de fréquences optiques ainsi généré, est composé de 21 modes, ayant tous une largeur de raie optique intrinsèque inférieure à 7 kHz, dont 9 modes sont sub-kHz. Pour un temps d’observation du spectre optique de 10 ms, ces modes exhibent tous une largeur de raie d’environ 37 kHz. Nous démontrons ainsi l’impact de la gigue d’impulsions sur les performances de la technique et nous soulignons l’intérêt d’une telle cohérence, pour le domaine des télécommunications optiques cohérentes (transmissions WDM cohérentes de type m-QAM avec des constellations d’ordre élevé, compatibles avec des débits multi-Tbit/s par raie). En dernier lieu, nous abordons une seconde technique consistant à pré-stabiliser la fréquence d’un laser par asservissement en boucle fermée. Elle repose sur l’utilisation d’un interféromètre à fibre déséquilibré comme référence pour réduire le bruit de fréquence d’un laser, situé particulièrement en basses fréquences. Appliquée à un laser mono-fréquence, elle a permis de réduire son bruit de fréquence technique conduisant ainsi à une nette amélioration de sa largeur de raie intégrée sur 3 ms, de 224 kHz à 37 kHz. Ce premier résultat représente un bon support vers l’exploration du potentiel des diodes à blocage de modes pour des applications métrologiques<br>Optical frequency combs obtained from mode-locked laser diodes are potential candidates for WDM networks. However, their lines exhibit usually a broad optical linewidth ( 1-100 MHz). Thus their use is incompatible for high order modulation formats WDM based systems. This thesis investigates one solution to overcome this limitation. It consists of using a feed-forward heterodyne technique to reduce the frequency noise of each comb-line and consequently their optical linewidths. In a first approach, the technique is applied to a single-mode laser. This allowed us to validate its proper working and to identify the intrinsic limits of the experimental device set up. The latter analysis enabled us to reveal that the minimum achievable frequency noise level by our system, corresponds to a 50 Hz intrinsic optical linewidth spectrum and a 1,6 kHz optical linewidth based on 10 ms observation time. This technique is then applied to an actively mode-locked laser diode demonstrating, at our system output, a 21-line optical frequency comb with intrinsic optical linewidths reduced to below 7 kHz. It is worth noting that 9 among them, exhibit sub-kHz linewidths. For an observation time of 10 ms, all lines share the same optical linewidth, almost equal to 37 kHz. We thus show that the timing jitter impacts the technique performances. We also highlight the relevance of such coherence level for coherent optical communication. Lastly, we study a laser frequency pre-stabilization technique based on a locking to an unbalanced fiber interferometer. When applied to a single-mode laser, the technique showed a reduction of its technical frequency noise, thus leading to a clear improvement of its integrated optical linewidth from 224 kHz to 37 kHz for 3 ms observation time. This first result provides a good support towards the exploration of mode-locked laser diodes potential for metrological applications
10
Fong, Bo-Jyun, and 馮柏鈞. "Relative Carrier-Envelope-Offset Frequency Locking between Ultrafast Two-Color Fiber Laser Systems by Feed-Forward Method." Thesis, 2014. http://ndltd.ncl.edu.tw/handle/71951547407122969401.
Full textAbstract:
碩士<br>輔仁大學<br>物理學系碩士班<br>102<br>In this thesis, we investigate the possible method to lock the relative carrier-envelope-offset (CEO) frequency of the ultrafast two-color fiber laser system with a synchronized pulse repetition frequency. First of all, the locking of the pulse repetition frequencies is achieved by the hybrid synchronization technique. After achieving the synchronization, the relative CEO frequency is obtained from the beat notes between the output of the Yb-fiber laser system and the supercontinuum (SC) generated by the Er-fiber laser system. We use the feed-forward method based on an acousto-optic frequency shifter (AOFS) to lock the relative CEO frequency at zero or a fixed frequency. When the relative CEO frequency is locked to zero, interference fringes between the SC of the Er-fiber laser system and the Yr-fiber laser output diffracted by the AOFS can be observed obviously in the optical spectrum near 1 μm. This means not only the CEO frequency but the CEO phase is stabilized.
Books on the topic "Feed forward layer"
1
Herreros, Ivan. Learning and control. Oxford University Press, 2018. http://dx.doi.org/10.1093/oso/9780199674923.003.0026.
Full textAbstract:
This chapter discusses basic concepts from control theory and machine learning to facilitate a formal understanding of animal learning and motor control. It first distinguishes between feedback and feed-forward control strategies, and later introduces the classification of machine learning applications into supervised, unsupervised, and reinforcement learning problems. Next, it links these concepts with their counterparts in the domain of the psychology of animal learning, highlighting the analogies between supervised learning and classical conditioning, reinforcement learning and operant conditioning, and between unsupervised and perceptual learning. Additionally, it interprets innate and acquired actions from the standpoint of feedback vs anticipatory and adaptive control. Finally, it argues how this framework of translating knowledge between formal and biological disciplines can serve us to not only structure and advance our understanding of brain function but also enrich engineering solutions at the level of robot learning and control with insights coming from biology.
2
Tisenhusen, Isabella Barbara. How to Win: The Future of Law with Artificial Intelligence. Bloomsbury Publishing Plc, 2025. https://doi.org/10.5040/9781526532770.
Full textAbstract:
We are living in a legal renaissance. Lawyers have never been more empowered to reimagine their profession thanks to AI. This title answers questions such as ‘Will AI be better than me at my job?’, ‘Why do I need to start using AI?’ and ‘How will AI enhance my capability as a legal professional?’ by providing advice on how lawyers and law firms can integrate new technology and improve their legal practice. The title further explains how to make interactions between lawyers and clients a win-win for all parties involved, while increasing client retention and profit for the firm. This book is written by a lawyer, for other lawyers, minus the legalese. It is based on the author’s personal observations of the legal industry and interviews with her colleagues across various practice areas and all levels of seniority. It aims to bring forward ideas for improvement. Using examples from the UK, EU and the US across corporate law, contract law, mergers and acquisitions and litigation, this title is relevant to lawyers in private practice specialising in all areas of the law. Are you a practising lawyer at a law firm? Do you sometimes feel like the legal industry is held back by tradition and could use some shaking up? Perhaps you are a mover and shaker yourself. In any case, this book is for you if you are curious about possibilities to improve the legal industry. The book also includes a list of AI tools that are beneficial in legal work. This title is included in Bloomsbury Professional's Cyber Law online service.
3
Stitzlein, Sarah M. Learning How to Hope. Oxford University Press, 2019. http://dx.doi.org/10.1093/oso/9780190062651.001.0001.
Full textAbstract:
Democracy is struggling in America. Citizens increasingly feel cynical about our system and doubt they can influence public policy. Distrustful of other Americans and elected officials, some are even turning to authoritarian alternatives. Hyperpartisanship and recent contentious presidential elections have deepened political despair. While some citizens get swept up in optimism during campaign cycles, they often later find themselves frustrated with elected leaders as they wait for change. This book seeks to revive democracy by teaching citizens how to hope. Hope animates life in a democracy, moving citizens forward through new challenges, new ideas, and new experiments. The form of hope described in this book is more than just a campaign slogan or a self-help program, it is an informed call to citizen engagement that opens new possibilities for our country. Drawing on examples from life in America today and pragmatist philosophy, this book explains how schools can cultivate hope through our habits and how action in our communities can sustain hope. It shows how we can build trust, grow political agency, and shape an improved American identity through hoping together. This book provides guidance for learning how to hope in schools, universities, and civil society. It describes what hope is, why it matters to democracy, and how to teach it.
4
Sommers, Joseph Michael, and Kyle Eveleth, eds. The Artistry of Neil Gaiman. University Press of Mississippi, 2019. http://dx.doi.org/10.14325/mississippi/9781496821645.001.0001.
Full textAbstract:
Neil Gaiman (1960-present) currently reigns in the literary world as one of the most critically-decorated and popular authors of the last fifty years. Perhaps best known as the writer of the Harvey, Eisner, and World Fantasy-award winning DC/ Vertigo series, The Sandman, Gaiman quickly became equally-renowned in literary circles for works such as Neverwhere, Coraline, the Hugo, Nebula, Locus, etc. award-winning American Gods, as well as the Newbery and Carnegie Medal-winning The Graveyard Book. For adults, for children, for the comic reader to the viewer of the BBC's Doctor Who, Gaiman's writing has crossed the borders of virtually all media and every language making him a celebrity on a world-wide scale. Despite Gaiman's incredible contributions to multiple national comics traditions (from such works as Miracleman to the aforementioned The Sandman), to the maturation of American comics as a serious storytelling medium, and to changing the rights of creators to retain ownership of their works, his work continues to be underrepresented in sustained fashion in comics studies. As American Gods tops ratings charts for Starz, Anansi Boys can be found in radio play from the BBC, and adaptations of some of his work from Trigger Warning and Fragile Things become standalone comics by renowned artists, it seems timely to bring the bulk of Gaiman's comics into the scholarly discussion. The thirteen essays and two interviews with Gaiman and his frequent collaborator, artist P. Craig Russell, a formal introduction, forward, and afterword examine the work (specifically-comics, graphic novels, picture books, visual adaptations of prose works, etc.) of Gaiman and a multitude of his collaborative illustrators. The essays radiate from an examination of Gaiman's work surrounding proclamations challenging his readers to "make good art'; what makes Gaiman's work unique and worthy of study lies in his eschewing of typical categorizations and typologies, his constant efforts to make good art-whatever form that art may take-howsoever the genres and audiences may slip into one another. What emerges is a complicated picture of a man who always seems fully-assembled virtually from the start of his career, but only came to feel comfortable in his own skin and his own voice far later in his life.
5
Dube, Opha Pauline. Climate Policy and Governance across Africa. Oxford University Press, 2016. http://dx.doi.org/10.1093/acrefore/9780190228620.013.605.
Full textAbstract:
This is an advance summary of a forthcoming article in the Oxford Research Encyclopedia of Climate Science. Please check back later for the full article.Africa, a continent with the largest number of countries falling under the category of Least Developed Countries (LDCs), remains highly dependent on rain-fed agriculture that suffers from low intake of water, exacerbating the vulnerability to climate variability and anthropogenic climate change. The increasing frequency and severity of climate extremes impose major strains on the economies of these countries. The loss of livelihoods due to interaction of climate change with existing stressors is elevating internal and cross-border migration. The continent is experiencing rapid urbanization, and its cities represent the most vulnerable locations to climate change due in part to incapacitated local governance. Overall, the institutional capacity to coordinate, regulate, and facilitate development in Africa is weak. The general public is less empowered to hold government accountable. The rule of law, media, and other watchdog organizations, and systems of checks and balances are constrained in different ways, contributing to poor governance and resulting in low capacity to respond to climate risks.As a result, climate policy and governance are inseparable in Africa, and capacitating the government is as essential as establishing climate policy. With the highest level of vulnerability to climate change compared with the rest of the world, governance in Africa is pivotal in crafting and implementing viable climate policies.It is indisputable that African climate policy should focus first and foremost on adaptation to climate change. It is pertinent, therefore, to assess Africa’s governance ability to identify and address the continent’s needs for adaptation. One key aspect of effective climate policy is access to up-to-date and contextually relevant information that encompasses indigenous knowledge. African countries have endeavored to meet international requirements for reports such as the National Communications on Climate Change Impacts and Vulnerabilities and the National Adaptation Programmes of Action (NAPAs). However, the capacity to deliver on-time quality reports is lacking; also the implementation, in particular integration of adaptation plans into the overall development agenda, remains a challenge. There are a few successes, but overall adaptation operates mainly at project level. Furthermore, the capacity to access and effectively utilize availed international resources, such as extra funding or technology transfer, is limited in Africa.While the continent is an insignificant source of emissions on a global scale, a more forward looking climate policy would require integrating adaptation with mitigation to put in place a foundation for transformation of the development agenda, towards a low carbon driven economy. Such a futuristic approach calls for a comprehensive and robust climate policy governance that goes beyond climate to embrace the Sustainable Development Goals Agenda 2030. Both governance and climate policy in Africa will need to be viewed broadly, encompassing the process of globalization, which has paved the way to a new geological epoch, the Anthropocene. The question is, what should be the focus of climate policy and governance across Africa under the Anthropocene era?
Book chapters on the topic "Feed forward layer"
1
Annema, Anne-Johan. "Dynamics of Single Layer Nets." In Feed-Forward Neural Networks. Springer US, 1995. http://dx.doi.org/10.1007/978-1-4615-2337-6_3.
Full text
2
Annema, Anne-Johan. "Cost Functions for Two-Layer Neural Networks." In Feed-Forward Neural Networks. Springer US, 1995. http://dx.doi.org/10.1007/978-1-4615-2337-6_10.
Full text
3
Annema, Anne-Johan. "Discretization of Weight Adaptations in Single-Layer Nets." In Feed-Forward Neural Networks. Springer US, 1995. http://dx.doi.org/10.1007/978-1-4615-2337-6_7.
Full text
4
Lee, Tsu-Chang. "Multi-Layer Feed-Forward Networks." In Structure Level Adaptation for Artificial Neural Networks. Springer US, 1991. http://dx.doi.org/10.1007/978-1-4615-3954-4_3.
Full text
5
Annema, Anne-Johan. "Cross-talk in Single-Layer Feed-Forward Neural Networks." In Feed-Forward Neural Networks. Springer US, 1995. http://dx.doi.org/10.1007/978-1-4615-2337-6_5.
Full text
6
Annema, Anne-Johan. "Unipolar Input Signals in Single-Layer Feed-Forward Neural Networks." In Feed-Forward Neural Networks. Springer US, 1995. http://dx.doi.org/10.1007/978-1-4615-2337-6_4.
Full text
7
Annema, Anne-Johan. "Learning Behavior and Temporary Minima of Two-Layer Neural Networks." In Feed-Forward Neural Networks. Springer US, 1995. http://dx.doi.org/10.1007/978-1-4615-2337-6_8.
Full text
8
Annema, Anne-Johan. "Biases and Unipolar Input signals for Two-Layer Neural Networks." In Feed-Forward Neural Networks. Springer US, 1995. http://dx.doi.org/10.1007/978-1-4615-2337-6_9.
Full text
9
Annema, Anne-Johan. "Precision Requirements for Analog Weight Adaptation Circuitry for Single-Layer Nets." In Feed-Forward Neural Networks. Springer US, 1995. http://dx.doi.org/10.1007/978-1-4615-2337-6_6.
Full text
10
Lio, Wai Hou. "Feed-Forward Model Predictive Control Layer on Wind Turbines." In Springer Theses. Springer International Publishing, 2018. http://dx.doi.org/10.1007/978-3-319-75532-8_7.
Full textConference papers on the topic "Feed forward layer"
1
Anuradha, E., Indraja C, Indupriya P, Prem Sai S, and Bala Subramanyam E. "Efficient Fault Detection in Photovoltaic Systems using Single Layer Feed Forward Neural Networks." In 2025 3rd International Conference on Data Science and Information System (ICDSIS). IEEE, 2025. https://doi.org/10.1109/icdsis65355.2025.11070677.
Full text
2
Donglian, Gao, and Shan Rui. "TR-BFGS Algorithm for Multi-layer Feed-Forward Neural Networks." In Its Applications and Embedded Sys (CDEE). IEEE, 2010. http://dx.doi.org/10.1109/cdee.2010.38.
Full text
3
Wang, Ran, Sam Kwong, Qingshan Jiang, and Ka-Chun Wong. "Active Learning Based on Single-Hidden Layer Feed-Forward Neural Network." In 2015 IEEE International Conference on Systems, Man, and Cybernetics (SMC). IEEE, 2015. http://dx.doi.org/10.1109/smc.2015.377.
Full text
4
Wisitpongphan, Nawaporn, Watchareewan Jitsakul, and Duangporn Jieamumporn. "Travel Time Prediction Using Multi-layer Feed Forward Artificial Neural Network." In 2012 4th International Conference on Computational Intelligence, Communication Systems and Networks (CICSyN 2012). IEEE, 2012. http://dx.doi.org/10.1109/cicsyn.2012.67.
Full text
5
Dutt, Micheal, Vimala Nunavath, and Morten Goodwin. "A Multi-layer Feed Forward Neural Network Approach for Diagnosing Diabetes." In 2018 11th International Conference on Developments in eSystems Engineering (DeSE). IEEE, 2018. http://dx.doi.org/10.1109/dese.2018.00060.
Full text
6
Sahoo, Pradyumna K., Ramaprasad Panda, Prasanta K. Satpathy, and Subrata Paul. "Voltage stability monitoring based on Feed Forward and Layer Recurrent Neural Networks." In 2014 6th IEEE Power India International Conference (PIICON). IEEE, 2014. http://dx.doi.org/10.1109/34084poweri.2014.7117623.
Full text
7
Hegadi, Ravindra S., and Parshuram M. Kamble. "Recognition of Marathi Handwritten Numerals Using Multi-Layer Feed-Forward Neural Network." In 2014 World Congress on Computing and Communication Technologies (WCCCT). IEEE, 2014. http://dx.doi.org/10.1109/wccct.2014.9.
Full text
8
Sahoo, Pradyumna K., Ramaprasad Panda, Prasanta K. Satpathy, and Subrata Paul. "Voltage stability monitoring based on Feed Forward and Layer Recurrent Neural Networks." In 2014 6th IEEE Power India International Conference (PIICON). IEEE, 2014. http://dx.doi.org/10.1109/poweri.2014.7117623.
Full text
9
Altosaar, Toomas, and Einar Meister. "Speaker recognition experiments in Estonian using multi-layer feed-forward neural nets." In 4th European Conference on Speech Communication and Technology (Eurospeech 1995). ISCA, 1995. http://dx.doi.org/10.21437/eurospeech.1995-89.
Full text
10
Youssefi, Bahar, Mitra Mirhassani, and Jonathan Wu. "Efficient mixed-signal synapse multipliers for multi-layer feed-forward neural networks." In 2016 IEEE 59th International Midwest Symposium on Circuits and Systems (MWSCAS). IEEE, 2016. http://dx.doi.org/10.1109/mwscas.2016.7870144.
Full textReports on the topic "Feed forward layer"
1
Arhin, Stephen, Babin Manandhar, Hamdiat Baba Adam, and Adam Gatiba. Predicting Bus Travel Times in Washington, DC Using Artificial Neural Networks (ANNs). Mineta Transportation Institute, 2021. http://dx.doi.org/10.31979/mti.2021.1943.
Full textAbstract:
Washington, DC is ranked second among cities in terms of highest public transit commuters in the United States, with approximately 9% of the working population using the Washington Metropolitan Area Transit Authority (WMATA) Metrobuses to commute. Deducing accurate travel times of these metrobuses is an important task for transit authorities to provide reliable service to its patrons. This study, using Artificial Neural Networks (ANN), developed prediction models for transit buses to assist decision-makers to improve service quality and patronage. For this study, we used six months of Automatic Vehicle Location (AVL) and Automatic Passenger Counting (APC) data for six Washington Metropolitan Area Transit Authority (WMATA) bus routes operating in Washington, DC. We developed regression models and Artificial Neural Network (ANN) models for predicting travel times of buses for different peak periods (AM, Mid-Day and PM). Our analysis included variables such as number of served bus stops, length of route between bus stops, average number of passengers in the bus, average dwell time of buses, and number of intersections between bus stops. We obtained ANN models for travel times by using approximation technique incorporating two separate algorithms: Quasi-Newton and Levenberg-Marquardt. The training strategy for neural network models involved feed forward and errorback processes that minimized the generated errors. We also evaluated the models with a Comparison of the Normalized Squared Errors (NSE). From the results, we observed that the travel times of buses and the dwell times at bus stops generally increased over time of the day. We gathered travel time equations for buses for the AM, Mid-Day and PM Peaks. The lowest NSE for the AM, Mid-Day and PM Peak periods corresponded to training processes using Quasi-Newton algorithm, which had 3, 2 and 5 perceptron layers, respectively. These prediction models could be adapted by transit agencies to provide the patrons with accurate travel time information at bus stops or online.
2
MacFarlane, Andrew. 2021 medical student essay prize winner - A case of grief. Society for Academic Primary Care, 2021. http://dx.doi.org/10.37361/medstudessay.2021.1.1.
Full textAbstract:
As a student undertaking a Longitudinal Integrated Clerkship (LIC)1 based in a GP practice in a rural community in the North of Scotland, I have been lucky to be given responsibility and my own clinic lists. Every day I conduct consultations that change my practice: the challenge of clinically applying the theory I have studied, controlling a consultation and efficiently exploring a patient's problems, empathising with and empowering them to play a part in their own care2 – and most difficult I feel – dealing with the vast amount of uncertainty that medicine, and particularly primary care, presents to both clinician and patient. I initially consulted with a lady in her 60s who attended with her husband, complaining of severe lower back pain who was very difficult to assess due to her pain level. Her husband was understandably concerned about the degree of pain she was in. After assessment and discussion with one of the GPs, we agreed some pain relief and a physio assessment in the next few days would be a practical plan. The patient had one red flag, some leg weakness and numbness, which was her ‘normal’ on account of her multiple sclerosis. At the physio assessment a few days later, the physio felt things were worse and some urgent bloods were ordered, unfortunately finding raised cancer and inflammatory markers. A CT scan of the lung found widespread cancer, a later CT of the head after some developing some acute confusion found brain metastases, and a week and a half after presenting to me, the patient sadly died in hospital. While that was all impactful enough on me, it was the follow-up appointment with the husband who attended on the last triage slot of the evening two weeks later that I found completely altered my understanding of grief and the mourning of a loved one. The husband had asked to speak to a Andrew MacFarlane Year 3 ScotGEM Medical Student 2 doctor just to talk about what had happened to his wife. The GP decided that it would be better if he came into the practice - strictly he probably should have been consulted with over the phone due to coronavirus restrictions - but he was asked what he would prefer and he opted to come in. I sat in on the consultation, I had been helping with any examinations the triage doctor needed and I recognised that this was the husband of the lady I had seen a few weeks earlier. He came in and sat down, head lowered, hands fiddling with the zip on his jacket, trying to find what to say. The GP sat, turned so that they were opposite each other with no desk between them - I was seated off to the side, an onlooker, but acknowledged by the patient with a kind nod when he entered the room. The GP asked gently, “How are you doing?” and roughly 30 seconds passed (a long time in a conversation) before the patient spoke. “I just really miss her…” he whispered with great effort, “I don’t understand how this all happened.” Over the next 45 minutes, he spoke about his wife, how much pain she had been in, the rapid deterioration he witnessed, the cancer being found, and cruelly how she had passed away after he had gone home to get some rest after being by her bedside all day in the hospital. He talked about how they had met, how much he missed her, how empty the house felt without her, and asking himself and us how he was meant to move forward with his life. He had a lot of questions for us, and for himself. Had we missed anything – had he missed anything? The GP really just listened for almost the whole consultation, speaking to him gently, reassuring him that this wasn’t his or anyone’s fault. She stated that this was an awful time for him and that what he was feeling was entirely normal and something we will all universally go through. She emphasised that while it wasn’t helpful at the moment, that things would get better over time.3 He was really glad I was there – having shared a consultation with his wife and I – he thanked me emphatically even though I felt like I hadn’t really helped at all. After some tears, frequent moments of silence and a lot of questions, he left having gotten a lot off his chest. “You just have to listen to people, be there for them as they go through things, and answer their questions as best you can” urged my GP as we discussed the case when the patient left. Almost all family caregivers contact their GP with regards to grief and this consultation really made me realise how important an aspect of my practice it will be in the future.4 It has also made me reflect on the emphasis on undergraduate teaching around ‘breaking bad news’ to patients, but nothing taught about when patients are in the process of grieving further down the line.5 The skill Andrew MacFarlane Year 3 ScotGEM Medical Student 3 required to manage a grieving patient is not one limited to general practice. Patients may grieve the loss of function from acute trauma through to chronic illness in all specialties of medicine - in addition to ‘traditional’ grief from loss of family or friends.6 There wasn’t anything ‘medical’ in the consultation, but I came away from it with a real sense of purpose as to why this career is such a privilege. We look after patients so they can spend as much quality time as they are given with their loved ones, and their loved ones are the ones we care for after they are gone. We as doctors are the constant, and we have to meet patients with compassion at their most difficult times – because it is as much a part of the job as the knowledge and the science – and it is the part of us that patients will remember long after they leave our clinic room. Word Count: 993 words References 1. ScotGEM MBChB - Subjects - University of St Andrews [Internet]. [cited 2021 Mar 27]. Available from: https://www.st-andrews.ac.uk/subjects/medicine/scotgem-mbchb/ 2. Shared decision making in realistic medicine: what works - gov.scot [Internet]. [cited 2021 Mar 27]. Available from: https://www.gov.scot/publications/works-support-promote-shared-decisionmaking-synthesis-recent-evidence/pages/1/ 3. Ghesquiere AR, Patel SR, Kaplan DB, Bruce ML. Primary care providers’ bereavement care practices: Recommendations for research directions. Int J Geriatr Psychiatry. 2014 Dec;29(12):1221–9. 4. Nielsen MK, Christensen K, Neergaard MA, Bidstrup PE, Guldin M-B. Grief symptoms and primary care use: a prospective study of family caregivers. BJGP Open [Internet]. 2020 Aug 1 [cited 2021 Mar 27];4(3). Available from: https://bjgpopen.org/content/4/3/bjgpopen20X101063 5. O’Connor M, Breen LJ. General Practitioners’ experiences of bereavement care and their educational support needs: a qualitative study. BMC Medical Education. 2014 Mar 27;14(1):59. 6. Sikstrom L, Saikaly R, Ferguson G, Mosher PJ, Bonato S, Soklaridis S. Being there: A scoping review of grief support training in medical education. PLOS ONE. 2019 Nov 27;14(11):e0224325.
3
Monetary Policy Report - January 2021. Banco de la República de Colombia, 2021. http://dx.doi.org/10.32468/inf-pol-mont-eng.tr1.-2021.
Full textAbstract:
Macroeconomic Summary Overall inflation (1.61%) and core inflation (excluding food and regulated items) (1.11%) both declined beyond the technical staff’s expectations in the fourth quarter of 2020. Year-end 2021 forecasts for both indicators were revised downward to 2.3% and 2.1%, respectively. Market inflation expectations also fell over this period and suggested inflation below the 3% target through the end of this year, rising to the target in 2022. Downward pressure on inflation was more significant in the fourth quarter than previously projected, indicating weak demand. Annual deceleration among the main groups of the consumer price index (CPI) was generalized and, except for foods, was greater than projected in the October report. The CPI for goods (excluding foods and regulated items) and the CPI for regulated items were subject to the largest decelerations and forecasting discrepancies. In the first case, this was due in part to a greater-than-expected effect on prices from the government’s “VAT-fee day” amid weak demand, and from the extension of some price relief measures. For regulated items, the deceleration was caused in part by unanticipated declines in some utility prices. Annual change in the CPI for services continued to decline as a result of the performance of those services that were not subject to price relief measures, in particular. Although some of the overall decline in inflation is expected to be temporary and reverse course in the second quarter of 2021, various sources of downward pressure on inflation have become more acute and will likely remain into next year. These include ample excesses in capacity, as suggested by the continued and greater-than-expected deceleration in core inflation indicators and in the CPI for services excluding price relief measures. This dynamic is also suggested by the minimal transmission of accumulated depreciation of the peso on domestic prices. Although excess capacity should fall in 2021, the decline will likely be slower than projected in the October report amid additional restrictions on mobility due to a recent acceleration of growth in COVID-19 cases. An additional factor is that low inflation registered at the end of 2020 will likely be reflected in low price adjustments on certain indexed services with significant weight in the CPI, including real estate rentals and some utilities. These factors should keep inflation below the target and lower than estimates from the previous report on the forecast horizon. Inflation is expected to continue to decline to levels near 1% in March, later increasing to 2.3% at the end of 2021 and 2.7% at year-end 2022 (Graph 1.1). According to the Bank’s most recent survey, market analysts expect inflation of 2.7% and 3.1% in December 2021 and 2022, respectively. Expected inflation derived from government bonds was 2% for year-end 2021, while expected inflation based on bonds one year forward from that date (FBEI 1-1 2022) was 3.2%.