As deep learning continues to push the boundaries with applications previously thought impossible, it has become more important than ever to reduce the associated resource costs. Data is not always abundant, labelling costs valuable human time, and deep models are demanding of computer hardware. In this dissertation, I will examine questions of minimalism in deep learning. I will show that deep learning can learn with fewer measurements, fewer weights, and less information. With minimalism, deep learning can become even more ubiquitous, succeeding in more applications and on more everyday devices.

Prediction of initial velocity of artillery system is a feature that is hard to determine with statistical and analytical models. Machine learning is therefore to be tested, in order to achieve a higher accuracy than the current method (baseline). An ensemble approach will be explored in this paper, centered around feed forward deep neural network and random forest regression. Furthermore, collinearity of features and their importance will be investigated. The impact of the measured error on the range of the projectile will also be derived by finding a numerical solution with Newton Raphsons method. For the five systemstest data was used on, the mean absolute errors were 26, 9.33, 8.72 and 9.06 for deep neural networks,random forest regression, ensemble learning and conventional method, respectively. For future works,more models should be tested with ensemble learning, as well as investigation on the feature space for the input data.

Indiana University-Purdue University Indianapolis (IUPUI) / Textual emotion classification is a task in affective AI that branches from sentiment analysis and focuses on identifying emotions expressed in a given text excerpt. It has a wide variety of applications that improve human-computer interactions, particularly to empower computers to understand subjective human language better. Significant research has been done on this task, but very little of that research leverages one of the most emotion-bearing symbols we have used in modern communication: Emojis. In this thesis, we propose several transformer-based models for emotion classification that processes emojis as input tokens and leverages pretrained models and uses them , a model that processes Emojis as textual inputs and leverages DeepMoji to generate affective feature vectors used as reference when aggregating different modalities of text encoding. To evaluate ReferEmo, we experimented on the SemEval 2018 and GoEmotions datasets, two benchmark datasets for emotion classification, and achieved competitive performance compared to state-of-the-art models tested on these datasets. Notably, our model performs better on the underrepresented classes of each dataset.

No description available.

No description available.

No description available.

No description available.

Equations of motion with required boundary conditions for doubly curved deep and thick composite shells are shown using two formulations. The first is based upon the formulation that was presented initially by Rath and Das (1973, J. Sound and Vib.) and followed by Reddy (1984, J. Engng. Mech. ASCE). In this formulation, plate stiffness parameters are used for thick shells, which reduced the equations to those applicable for shallow shells. This formulation is widely used but its accuracy has not been completely tested. The second formulation is based upon that of Qatu (1995, Compos. Press. Vessl. Indust.; 1999, Int. J. Solids Struct.). In this formulation, the stiffness parameters are calculated by using exact integration of the stress resultant equations. In addition, Qatu considered the radius of twist in his formulation. In both formulations, first order polynomials for in-plane displacements in the z-direction are utilized allowing for the inclusion of shear deformation and rotary inertia effects (first order shear deformation theory or FSDT). Also, FSDTQ has been modified in this dissertation using the radii of each laminate instead of using the radii of mid-plane in the moment of inertias and stress resultants equations. Exact static and free vibration solutions for isotropic and symmetric and anti-symmetric cross-ply cylindrical shells for different length-to-thickness and length-to-radius ratios are obtained using the above theories. Finally, the equations of motion are put together with the equations of stress resultants to arrive at a system of seventeen first-order differential equations. These equations are solved numerically with the aid of General Differential Quadrature (GDQ) method for isotropic, cross-ply, angle-ply and general lay-up cylindrical shells with different boundary conditions using the above mentioned theories. Results obtained using all three theories (FSDT, FSDTQ and modified FSDTQ) are compared with the results available in literature and those obtained using a three-dimensional (3D) analysis. The latter (3D) is used here mainly to test the accuracy of the shell theories presented here.

Lesbians have historically lived in obscurity and isolation because living outwardly as a lesbian carried with it the almost certain loss of social standing, family, and friends (Blando, 2001). For lesbians who grew up in the Deep South, isolation and the pressure to conform was greater than anywhere in the United States (Barton, 2010). Most Deep Southerners were homophobic, especially in rural areas where people were deeply religious and had little exposure to sexual minorities. The researcher used a qualitative phenomenological approach to explore the meaning and significance of growing up lesbian in the rural Deep South. The sample included 12 Caucasian lesbians, ages 45 to 62. Four clusters of themes emerged from the interviews. Those clusters were: (1) emerging sexuality, (2) the mark of fatal difference, (3) denial of lesbian identity, and (4) conforming to Deep Southern social mores. Themes within those clusters described how delays in both lesbian identity development (Cass, 1984) and psychosocial development (Erikson, 1975) occurred in each of the participants because of the intensely religious and homophobic environments in which they were raised. Denunciation of participants' personal identities began with the first expressions of their sexual identities in elementary school. Ridiculed at a young age because of attractions to girls, participants cycled back through developmental crises involving shame, doubt, and inferiority. They entered adolescence disturbed about their developing sexualities, to discover that parents and faith-based communities were homophobic. Therefore, at the time when participants faced the most critical developmental crises of their lives (Erikson, 1975), they feared rejection by their parents, communities, and God. Participants sought to suppress or deny their lesbianism. Suppression of lesbian identity came with emotional and developmental costs, including substance abuse, unwanted marriages, and role confusion. Unable to find needed resources and role models, participants conformed to the social mores of the rural community for periods ranging from five to twenty years. Eventually, each participant in this study left her rural origins to begin claiming her lesbian identity. Retrospectively, each woman recognized that in the era in which they grew up, communities in the rural Deep South demanded conformity and resisted allowing members to individuate. Thus, participants in this study entered adulthood, and sometimes middle age, with a number of unresolved developmental crises, particularly as those crises related to sexual orientation.

A rigorous first order shear deformation theory (FSDT) is employed along with modified ABD parameters to analyze static and free vibration behavior of generally laminated beams and shafts. Different approaches for calculating composite beam stiffness parameters have been considered and the most accurate one that accounts for material couplings have been used to analyze static and free vibration behaviors of straight beams with different laminates and boundary conditions. In order to analyze curved beams, the term (1+z/R) is exactly integrated into ABD parameters formulation and an equivalent modulus of elasticity is used instead of traditional stiffness terms to account for both the deepness and material coupling of the beam structures. The model has been solved analytically for simply supported boundary conditions and the general differential quadrature (GDQ) technique has been used for other boundary conditions. The results for deflection, moment resultants, and natural frequencies of straight and curved beams with different deepness ratio (often called depth ratio), slenderness ratio, lamination, and boundary conditions are compared with those obtained from accurate three dimensional finite element simulations using ANSYS. The results were in close proximity to three dimensional finite element results. The model is then applied to transverse vibration analysis of multi-span generally laminated composite shafts with a lumped mass using GDQ. The results for natural frequencies are compared to experimental and other analytical models as well as finite element simulation. The results in the present analyses were found accurate. Conclusively, it has been shown that when considering more accurate stiffness parameters, a First Order Shear Deformation Theory can accurately predict static and free vibration behaviors of composite beams and multispan shafts of any deepness, lamination and boundary conditions.