Search

Deep neural networks have provided a computational framework for understanding object recognition, grounded in the neurophysiology of the primate ventral stream, but fail to account for how we process relational aspects of a scene. For example, deep neural networks fail at problems that involve enumerating the number of elements in an array, a problem that in humans relies on parietal cortex. Here, we build a 'dual-stream' neural network model which, equipped with both dorsal and ventral streams, can generalise its counting ability to wholly novel items ('zero-shot' counting). In doing so, it forms spatial response fields and lognormal number codes that resemble those observed in macaque posterior parietal cortex. We use the dual-stream network to make successful predictions about behavioural studies of the human gaze during similar counting tasks.

GaSe is an important member of the post-transition metal chalcogenide family and is an emerging two-dimensional (2D) semiconductor material. Because it is a van der Waals material, it can be fabricated into atomic-scale ultrathin films, making it suitable for the preparation of compact, heterostructure devices. In addition, GaSe possesses unusual optical and electronic properties, such as a shift from an indirect-bandgap single-layer film to a direct-bandgap bulk material, rare intrinsic p-type conduction, and nonlinear optical behaviors. These properties make GaSe an appealing candidate for the fabrication of field-effect transistors, photodetectors, and photovoltaics. However, the wafer-scale production of pure GaSe single crystal thin films remains challenging. This study develops an approach for the direct growth of nanometer-thick GaSe films on GaAs substrates using molecular beam epitaxy. It yields smooth thin GaSe films with the rare {\gamma}'-polymorph. We analyze the formation mechanism of {\gamma}'-GaSe using density functional theory and speculate that it is stabilized by Ga vacancies since the formation enthalpy of {\gamma}'-GaSe tends to become lower than that of other polymorphs when the Ga vacancy concentration increases. Finally, we investigate the growth conditions of GaSe, providing valuable insights for exploring 2D/3D quasi-van der Waals epitaxial growth.

GaAs(111)B is a semiconductor substrate widely used in research and commercial fields due to its low cost, mature synthesis technology, and excellent properties for manufacturing electronic devices. It is not only used to grow three-dimensional (3D) strongly-bonded materials, but has also been used as a substrate for layered, van der Waals (vdW)-bonded chalcogenide film growth. However, GaAs(111)B wafers cannot be directly used for growing epitaxial vdW chalcogenide films for two reasons: (1) the GaAs surface has a substantial number of dangling bonds that need to be passivated for vdW layers growth; (2) the substrate surface is covered with a thin epi-ready oxide layer which must be removed before film growth. In this paper, we optimize the method for deoxidizing GaAs(111)B substrates under a Se overpressure and successfully create a smooth, deoxidized, and passivated substrate for subsequent growth of vdW chalcogenide materials. We demonstrate the benefits of this method for the growth of vdW chalcogenide thin films using GaSe as a representative of vdW chalcogenides. In addition, we find that severely aged substrates have difficulty maintaining a smooth surface during the deoxidation and passivation process and cause GaSe crystals to nucleate in random shapes and orientations. We describe a method using water droplet testing to determine the age of the substrate. Finally, X-ray photoelectron spectroscopy (XPS) characterization reveals that the natural aging of GaAs(111)B in the air results in an increase in surface oxides, Ga2O3 and As2O3, while exposure to ultraviolet (UV)-ozone not only enhances the contents of these two oxides but also generates a new oxide, As2O5. Our research contributes to expanding the compatibility of GaAs(111)B with diverse growth materials and the production of high-quality heterostructure devices.

The introduction of superconductivity to the Dirac surface states of a topological insulator leads to a topological superconductor, which may support topological quantum computing through Majorana zero modes. The development of a scalable material platform is key to the realization of topological quantum computing. Here we report on the growth and properties of high-quality (Bi,Sb)2Te3/graphene/gallium heterostructures. Our synthetic approach enables atomically sharp layers at both hetero-interfaces, which in turn promotes proximity-induced superconductivity that originates in the gallium film. A lithography-free, van der Waals tunnel junction is developed to perform transport tunneling spectroscopy. We find a robust, proximity-induced superconducting gap formed in the Dirac surface states in 5-10 quintuple-layer (Bi,Sb)2Te3/graphene/gallium heterostructures. The presence of a single Abrikosov vortex, where the Majorana zero modes are expected to reside, manifests in discrete conductance changes. The present material platform opens up opportunities for understanding and harnessing the application potential of topological superconductivity., Comment: 19 pages, 4 figures, with a Supplemental Information. Comments are welcome

In this perspectives brief, authors James Kvaal and Jessica Thompson explore the challenge of college affordability and summarize the campaign proposals to address it. Driven by steadily rising college costs and student debt, the 2020 presidential campaign has put the issues of college costs and student debt on the agenda as never before. Many candidates are promising to transform the federal investment in college affordability, but there is great variety in how they would structure their initiatives. The debate on the strengths and weaknesses of these plans on the campaign trail is likely to have a substantial influence on future higher education policy.

For decades, state disinvestment in public higher education, paired with inequitable funding across institution types, has undermined states' ability to provide accessible and affordable higher education opportunities. In recent years, varied higher education stakeholders began calling for an end to state disinvestment through a reimagined federal-state partnership to better support public colleges. The authors of this report are specifically proposing a framework for a new federal-state partnership that will equip states to maintain their investment in public colleges across economic cycles. The framework in this report outlines three key design principles that a federal-state partnership must include to be effective: (1) providing strong incentives to maintain and increase state investment; (2) ensuring stable funding across economic cycles; and (3) making headway in closing racial and economic equity gaps in access and attainment. To account for changing budget realities, the report outlines several adjustable policy levers that Congress could scale up or down depending on available funding, as opposed to proposing a minimum or recommended dollar amount for federal spending. The authors also provide illustrated examples at several levels of funding.

We give an approximation algorithm for MaxCut and provide guarantees on the average fraction of edges cut on $d$-regular graphs of girth $\geq 2k$. For every $d \geq 3$ and $k \geq 4$, our approximation guarantees are better than those of all other classical and quantum algorithms known to the authors. Our algorithm constructs an explicit vector solution to the standard semidefinite relaxation of MaxCut and applies hyperplane rounding. It may be viewed as a simplification of the previously best known technique, which approximates Gaussian wave processes on the infinite $d$-regular tree., Comment: 1+9 pages

This new TICAS (The Institute for College Access & Success) report details existing requirements that govern the calculation and provision of post-college employment metrics, documents weaknesses in existing data, and recommends federal and state policy changes to support comparable, accurate, accessible information for measuring post-college employment outcomes.

This new TICAS (The Institute for College Access & Success) brief explores how the punitive consequences of defaulting on student debt can perversely make it harder for struggling borrowers to return to school and regain their financial footing, even though graduating from college will make it easier to repay their loans. The brief includes recommendations for how policymakers and colleges can help borrowers better prevent and get out of default.

Centered Kernel Alignment (CKA) was recently proposed as a similarity metric for comparing activation patterns in deep networks. Here we experiment with the modified RV-coefficient (RV2), which has very similar properties as CKA while being less sensitive to dataset size. We compare the representations of networks that received varying amounts of training on different layers: a standard trained network (all parameters updated at every step), a freeze trained network (layers gradually frozen during training), random networks (only some layers trained), and a completely untrained network. We found that RV2 was able to recover expected similarity patterns and provide interpretable similarity matrices that suggested hypotheses about how representations are affected by different training recipes. We propose that the superior performance achieved by freeze training can be attributed to representational differences in the penultimate layer. Our comparisons of random networks suggest that the inputs and targets serve as anchors on the representations in the lowest and highest layers., Comment: 4 pages, 4 figures, Conference on Cognitive Computational Neuroscience 2019